Has decentralization contributed to democratization at the local level in Mali? A comparative study of three local governments: Kontela, Mopti and Tambaga

Siri Bjerkreim Hellevik

Hovedoppgave i statsvitenskap, Institutt for Statsvitenskap

UNIVERSITETET I OSLO

October 2004

i

Contents

MAPS OF MALI AND THE THREE CASES v

ACKNOWLEDGEMENTS vii

ABBREVIATIONS AND EXPRESSIONS ix

1. INTRODUCTION TO THE STUDY OF LOCAL GOVERNMENT IN MALI...... 1

1.1 DECENTRALIZATION ...... 2

1.2 DEMOCRACY AND DEMOCRATIZATION...... 5

1.3 STATING THE PROBLEM - LINKING DECENTRALIZATION AND DEMOCRATIZATION...... 6

1.4 AN OUTLINE OF THE THESIS...... 8

1.5 CONCLUSION...... 10

2. FRAMEWORKS AND THEORETICAL EXPLANATIONS ...... 11

2.1 GENERAL PRESENTATION OF ASSESSMENT FRAMEWORKS ...... 11

2.2 GENERAL EXPLANATORY ARGUMENTS...... 12

2.3 SPECIFIC EXPLANATORY ARGUMENTS ...... 13

2.3.1 Situating the theoretical arguments in the theoretical debate ...... 13

2.3.2 The specific explanatory arguments of the three theories ...... 15

2.4 CONCLUSION...... 16

3. APPROACH AND METHOD ...... 17

3.1 THE CRITICAL CASE OF MALI AND THE SUB-CASES...... 17

3.2 RESEARCH DESIGNS ...... 20

3.3 TRIANGULATION, SOURCES AND FIELDWORK ...... 21

3.4 METHODOLOGICAL CHALLENGES...... 22

3.5 CONCLUSION...... 24

4. THE DECENTRALIZATION REFORM IN MALI ...... 25 ii

4.1 QUESTIONS ON THE DECENTRALIZATION REFORM ...... 25

4.2 THE INITIATIVE TO DECENTRALIZE - ACTORS AND STRATEGIES ...... 26

4.2.1 The participants at the National Conference ...... 26

4.2.2 National government and political parties...... 26

4.2.3 The role of the state administration ...... 27

4.3 ELEMENTS AND PROCESSES OF THE REFORM ...... 27

4.3.1 MDD and legislation...... 27

4.3.2 Participation by the population ...... 28

4.3.3 Establishing new institutions and devolving power ...... 29

4.3.4 State assistance and local revenue base ...... 30

4.4 CHALLENGES TO THE REFORM-CONTEXTUAL FACTORS ...... 31

4.5 CONCLUSION ...... 33

5. OPERATIONALIZATION OF ANALYTICAL FRAMEWORKS ...... 34

5.1 ASSESSING AND OPERATIONALIZING SUBSTANTIAL DEMOCRATIZATION ...... 34

5.2 ASSESSING AND OPERATIONALIZING DEMOCRATIC DECENTRALIZATION ...... 39

5.3 CONCLUSION ...... 41

6. SYNOPSIS OF THREE LOCAL GOVERNMENTS ...... 42

6.1 THE CASE OF KONTELA ...... 42

6.2 THE CASE OF TAMBAGA ...... 43

6.3 THE CASE OF MOPTI ...... 44

6.4 CONCLUSION ...... 45

7. ASSESSMENT OF SUBSTANTIAL DEMOCRATIZATION IN THREE LOCAL GOVERNMENTS...... 46

7.1 STATE OF AFFAIRS OF INSTITUTIONS AND RIGHTS...... 47 iii

7.1.1 State administration...... 47

7.1.2 Local Government- Municipal Council and Administration ...... 48

7.1.3 Accountability ...... 52

7.1.4 Free and fair elections?...... 58

7.1.5 Civil and political rights, citizenship and the judiciary...... 59

7.1.6 Civil society...... 60

7.2 ACTORS’ CAPACITY TO USE INSTITUTIONS AND RIGHTS...... 63

7.3 CONDITIONING FACTORS FOR ACTORS’ CAPACITY...... 70

7.4 CONCLUSION...... 75

8. GENERAL EXPLANATIONS AND THEORETICAL ARGUMENTS ...... 77

8.1 GENERAL EXPLANATIONS FOR THE CASES’ OUTCOME ...... 77

8.2 THE THREE SPECIFIC THEORETICAL APPROACHES...... 79

8.2.1 Political instrumentalization of disorder? ...... 79

8.2.2 A bifurcated state? ...... 81

8.2.3 Local politics- a complex game ...... 82

8.3 CONCLUSION...... 83

9. ASSESSING THE THEORETICAL ARGUMENTS ...... 84

9.1 A BIFURCATED STATE?...... 84

9.2 POLITICAL INSTRUMENTALIZATION OF DISORDER? ...... 89

9.3 LOCAL POLITICS - A COMPLEX GAME...... 99

9.4 COMPARISON WITHIN CASES ...... 102

9.4.1 Kontela and Tambaga...... 102

9.4.2 Mopti...... 105

9.5 GENERALIZATION OF FINDINGS ...... 108 iv

9.6 CONCLUSION ...... 112

10. CONCLUDING REMARKS ...... 114

10.1 MAIN FINDINGS, EXPLANATIONS AND CONCLUSIONS...... 114

10.2 EXPERIENCES FROM OTHER COUNTRIES ...... 124

10.3 FUTURE ISSUES OF RESEARCH ...... 126

10.4 FINAL CONCLUSION...... 126

REFERENCES………………………………………………………………………………………126

APPENDICES………………………………………………………………………………………. 137

1. TITLES OF INFORMANTS THAT ARE NOT ANONYMIZED

2. FRAMEWORK OF SUBSTANTIAL DEMOCRATIZATION

3. A SUMMARY OF THE ASSESSMENT OF POLITICAL ACTORS’ CAPACITY IN SECTION 7.2

FIGURES: 10.1 ASSESSMENT OF SUBSTANTIAL DEMOCRATIZATION……………………..114

v

Source: http://www.aircraft-charter-world.com/airports/africa/mali.htm

Source: Mission de Décentralisation (Mali) vi

Source: Mission de Décentralisation (Mali)

Source: Mission de Décentralisation (Mali)

vii

Acknowledgements

Many people have helped me during my work with this thesis. First of all, I would like to thank my main supervisor Olle Törnquist for inspiration and all his critical and challenging comments during the process of writing. Secondly, I would like to thank the co-supervisor Øivind Hetland for having contributed with in-depth contextual knowledge on Mali, good comments, and help with literature. Arild Schou who was my supervisor during the initial stages of the thesis also deserves a thanks. In addition to my supervisors, I am also in debt to Kristian Stokke, Kristin Tellefsen Engh, and the late Mariken Vaa for giving me fruitful remarks that have helped me improve the thesis. Further, thanks to Alida Boye at SUM for her enthusiasm and good inputs.

I am also grateful to the network of Local Politics in Developing countries for having provided me with a scholarship and lots of inspiration through the LPD research colloquiums. Through working for the LPD network based at SUM I have also benefited from discussions with researchers and students at SUM.

As to my fieldwork there are many people to thank. Thanks to Aud and Jon Bovim for giving me background knowledge and contacts in the Kayes region. Thanks to Eli Tveitane at MPN in Bamako for helping with transport and other practicalities before and during my fieldwork in the Kayes region. In Kontela I would most of all like to thank Karen Ekern for giving so much of her time to guide me around on bike to several informants, for introducing me to people wherever we went, as well as for giving me a place to stay. Without her the field work in Kontela would not have been possible to carry through. Further, thanks to all my informants for taking off time to answer my questions and giving me insight into politics in Kontela. In particular, I would like to thank Demba Diallo for valuable information during and after fieldwork, in conversation, e-mails and for documents he gave me.

In Tambaga, I am grateful to all my informants, in particular to N’Fali Dembele for interpretation and for giving me additional information after fieldwork. Moreover, viii thanks to Anita and Hølje Haugsjå for providing me with a place to stay, transport, information, and for introducing me to some people in the municipality.

The fieldwork in Mopti would not have gone smoothly without all the assistance and help from Boubacar Ba. Mamadou Balagny also deserves a thanks for all information during and after fieldwork. Furthermore, thanks to Gunn Gjeldstad at MELM for helping with transport and a place to stay during my field work in Mopti. In addition, the informants in Bamako and Bafoulabé are to be thanked for having answered all my questions.

I am also grateful to the Nordic Africa Institute for facilitating my literature study there, as well as for the grant I got for field work. Thanks to the Department of Political Science, University of Oslo and the Ministry of Defense for funding my field work as well.

In addition, thanks to Hanne, Hege, Lill, and Therese for having read parts of my thesis and giving me critical comments. Thanks to my parents for encouraging me in my work, and last, but most of all, many thanks to Petter for love, support, and patience during my work with this thesis.

ix

Abbreviations

ADEMA- Alliance pour la Démocratie au Mali

AIDEB- Appui Intégrale de Développement au cercle de Bafoulabé

ANICT-Agence Nationale d’Investissement des Collectivités Territoriales

BDIA- Bloque pour la Démocratie

CCC-Centre des Conseils Communaux

COPPO- Collectif des Parties d’Opposition

GIE- Groupement d’Interêt Economique (economic interest group)

IMF- International Monetary Fund

MDRI-Mission de Decentralisation et Reformes Institutionnels

PARENA- Parti de la renaissance national

PDP- Parti pour la démocratie et le progress

RPM- Rasemblement pour le Mali

UDD- Union pour la démocratie et le développement,

UNDP- United Nations Development Program

URD- Union Republique pour la Démocratie

US-RDA- Union Soudanaise-Rassemblement Démocratique Africain

WB-the World Bank

Expressions

Arrondissement- the former level of administration before the decentralization reform

Bambara- ethnic group

Customary power- the village/ward/fraction chief (customary used instead of “traditional”)

Diawambé- ethnic group

Fulani- ethnic group

Khassonké- ethnic group, primarily situated in the middle of the Kayes region

Malinké- ethnic group, primarily living in the southern part of the Kayes region x

1

1. Introduction to the study of local government in Mali During the wave of democratization in Africa of the early 90s, many states tried to implement democratic principles. Consequently, most African states are today “multiparty electoral regimes” (Van de Walle 2002:67). However, many of the states are still not free, thus not truly democratic (URL:www.freedomhouse.org). Therefore, since late 90s the call has turned to democratic decentralization, which involves an overall democratization of states by establishing local democratic structures with devolved authority (Johnson 2001). Many countries outside Africa are undertaking such decentralization reforms or have already done so. In Africa, the reforms are a reaction to many years of centralized rule and a rent-seeking administration, and have been initiated by international donors, opposition forces and national civil society (Olowu and Wunsch 2003). They all anticipate that decentralization will lead to local government where responsiveness and accountability exist, and where participation is enhanced because of the proximity between local government and the electorate. Furthermore, local governments are expected to be better informed than national governments about people’s needs. Moreover, greater knowledge of people’s needs, combined with the mechanism of accountability, is expected to result in higher achievements in development. Development organizations now fund and cooperate directly with local governments and local associations in order to achieve this goal and avoid corruption and inefficient bureaucracy. According to both the development organizations and decentralization proponents among scholars, decentralization is therefore a better instrument for achieving local development and democratization (WB 1997; UNDP 2002). In this thesis, I will concentrate on the decentralization debate concerning Africa, focusing on the reform in Mali. Mali is a landlocked nation in West Africa and is number 174 out of 177 on the Human Development Index of 2004 (UNDP 2004). It became independent from France in 1960, and had authoritarian rule until 1991, when the 2 democratic republic of Mali was established. It is on its way to consolidation of national democracy (Foumonyoh 2001).

1.1 Decentralization The problem which is addressed in this thesis is whether decentralization has contributed to democratization at the local level in Mali. The answer to this problem will be sought in case studies of the outcome of one reform at the local level, which assess the impact of decentralization in terms of whether democratization has begun to materialize or not. It is thus important to start by delineating the concepts decentralization and democratization. In this section, I will deal with decentralization, and in the following one with democratization. Decentralization is not a new concept or a new kind of reform. It has reappeared on the agenda in many countries as well as in international development organizations for at least the last thirty years. The theoretical and empirical debate on decentralization is voluminous, and people use the term in various ways. Some argue that decentralization implies devolution of authority (Villadsen 1999). Others believe that this is just one form of decentralization. One may separate the different types of decentralization from one another by measuring how much autonomy and how many tasks are transferred to the local level as a result of each type of decentralization (Sundstøl-Eriksen et al 1999). The main modes of decentralization are administrative, political, fiscal, and market decentralization. Administrative decentralization or de-concentration involves moving state administration to the local level, while maintaining authority at the centre. Devolution is equal to political decentralization, and implies that transfer of power takes place as well. Fiscal decentralization can take various forms, but it always includes “shifting some responsibilities for expenditures and/or revenues to lower levels of government”(URL:www1.worldbank.org (source b)). Market decentralization means transferring public services to private sector, including firms and civil society organizations, keeping responsibility at the central level (URL:www.fao.org). This may take the form of privatization or deregulation (ibid.). The main problem with market decentralization is that the entity delivering the services is not accountable to the citizens. In this thesis, I will mainly deal with 3 democratic decentralization, because this is the aim of the reform that I will study. A common definition of democratic decentralization is “meaningful authority devolved to local units of governance that are accessible and accountable to the local citizenry, who enjoy full political rights and liberty” (Blair 2000:21). Thus, both de- concentration and devolution are included in democratic decentralization. This thesis concentrates on the debate concerning decentralization in Africa, but some general contributions are briefly mentioned, too. Democratic decentralization first came about as development policy in the early 90s, though many earlier attempts at decentralizing, few of which were successful had been made after independence in Africa, mainly in the form of deconcentration (Olowu 2001:3, Ribot 2002). In general, democratic decentralization has been linked by international donors to the approaches of good governance and neo-liberalism, which appeared by mid 80s (Olowu et al 2004). At first, neo-liberalism stood for reduction of the state’s role in favor of privatization and market structures. The structural adjustment programs of the 80s followed this approach, and the World Bank (WB) and the International Monetary Fund (IMF) were the ones promoting them. These programs largely failed to facilitate development in Africa. In early 90s, the United Nations Development Fund (UNDP) reacted against the neo-liberal approach, because of the latter’s unique focus on privatization and reduction of the state‘s role. UNDP’s reaction was to give neo-liberalism “a human face”, which implied focusing on public institutions’ significance for development to succeed. Hence, good governance appeared as a development policy (see UNDP 2003 for definition). The WB and IMF and other international donors incorporated good governance in their policies (Harriss et al 2004:23; World Bank 1992:v). By late 90s, UNDP included participatory local democracy as a means to reach good governance, which involved emphasizing decentralization, social capital, and civil society (UNDP 2002). Organizations with a neo-liberal approach have adapted these ideas, but given them different meanings, depending on their own ideological standing. Further, there has been an odd consensus between neo-liberals and left-oriented activists on these issues (Harris et al 2004:1). Both groups have attached great importance to community involvement, emphasizing civil society‘s significant role in local development (ibid.). 4

This thesis will not give a detailed description of the overall debate on all these issues, because it is too extensive. It suffices to say that many proponents, such as international donors already mentioned, see decentralization as the best response to lack of participation and democratic practices, poverty, corrupt state administrations, and inefficient public service delivery. Hence, proponents of democratic decentralization have made local government the panacea of all these problems. Nevertheless, one of democratic decentralization’s major proponents, the WB recognizes that there are many obstacles to it (URL:www1.worldbank.org (source a)). Although democratic decentralization is promising in terms of achieving democratization in Africa, one may criticize proponents of the democratic decentralization approach presented here for not considering varying impact of local political context on such reforms. Proponents also tend to overestimate the central state’s authority at the local level when studying decentralization in Africa (Boone 1998:2). Even if ideas of international organizations are important in reforms, local processes and contexts, as well as power relations determine the implementation and outcome of reforms (Crook and Sverisson 2001; Therkildsen 2000). The neglect of these processes and relations by international donors and policy-makers makes it difficult for them to comprehend such reforms’ outcome. Consequently, one needs localized contextual analyses of decentralization reforms (Boone 1998:3), which is this thesis’ aim. The thesis will shed light on local processes and result of one reform, the Malian one. Hence, it does not suffice to outline Malian policies and their implementation at the local level, because this will not capture the entire local decentralization process. Rather, focus must be on formal and informal actors involved locally, as state, municipal council, and village chiefs. The thesis will study the interaction between these actors in the reform. It will look at the period from 1999, when municipal councils were elected, until the end of 2003 when fieldwork was done. It is however too encompassing to study all the reform’s elements. The specific aspects in focus follow from democratic decentralization’s promises: increased participation, an accountable and representative local government, and local development, which all may contribute to democratization. 5

The substantial democracy definition/framework (Törnquist 2002, 2004) assesses these factors, and is thus fruitful to employ in the thesis. The definition/framework concentrates on local institutions, as well as human rights, and includes assessing whether institutions are open, representative and accountable. It also measures people’s participation, through examining their capacity to use institutions and rights to further democracy. However, in order to study accountability thoroughly, the thesis has added the variables on accountability by Crook and Manor (1998) and Blair (2000).

1.2 Democracy and democratization The way one defines democracy influences the assessment of its existence and quality. There are numerous definitions of democracy, often set up as antithesis to each other (Beetham 1999:1). Still, “there is a wide agreement that the essence of democracy is ‘popular control of public affairs based on political equality’ ” (Törnquist 2004:301). The mainstream, minimalist definitions of liberal democracy include free and fair elections; some civil and political rights and majority rule (f ex. Schumpeter 1987). Some African scholars have criticized the liberal democracy definition for being irrelevant, because it is a Western concept and separated from democracy as such (Ake 2000; Tandon 1995; Mkandawire 1994, based on Reigstad 2002). Others argue that democracy is universal and not tied to the Western based model (Olokushi 1998). Some theorists also claim that there is a difference between liberal democracy and Western practice of democracy. Hence, democracy is a question of empirical outcome (Reigstad 2002:73). Recent contributions have a broader democracy definition, such as Beetham (1999), who includes all civil and political rights in his definition, while arguing that social, cultural and economic rights are in mutual relationship with democracy. This thesis employs another such encompassing definition: The substantial democracy definition (Törnquist 2002, 2003, 2004). It builds on Beetham (1999), but also includes people’s capacity to use democracy to further instrumental or democratic aims. This thesis’ emphasis is on democratization, because I study a reform process that is to create local democracy. Due to the reform’s recent origin (1999), it is too 6 early to study whether it has led to local democracy or not. Rather, it is fruitful to look at the process of establishing local democracy, i.e. whether elements of democratization have materialized. While democracy is a regime form, democratization is a continuing process of introducing democratic institutions and principles or deepening already existing ones (Törnquist 1999). Whether a regime is democratizing, depends on the definition of democracy employed (Törnquist 2002). The transition literature has recently been dominant in setting the agenda of the democratization debate (f.ex Lintz and Stephan 1996). Two of its contributors focusing on democratization in Africa are Bratton and Van de Walle (1997). Van de Walle (2001) argues that democratization results from “a crisis in the post-colonial neo-patrimonial order”. Gill (2002) criticizes him and others who belong to the transition “school” for neglecting civil society’ s crucial role in transitions to democracy. Further, Törnquist (1999 and 2004) criticizes them for their emphasis on elite-driven pacts instead of popular-based democratization by providing another framework as an alternative interpretation of democratization, elaborated on in chapters two and seven (see app. 2 in this thesis for figures).

1.3 Stating the problem - linking decentralization and democratization The problem at stake in this thesis is:

Has decentralization in Mali contributed to democratization at the local level? If so, how, and to what extent?

The thesis is delimited to look at the changes that have occurred since the implementation of the decentralization reform at the local level, i.e. the municipal level. In order to understand the reform’s result, it is fruitful to look at this level, which is the lowest level of governance, because here people encounter the new decentralized institutions. The unit of analysis is local government, i.e. municipal council and administration, but I will also examine local government’s interaction with other actors, such as already existing local power structures, as well as its 7 relation to the national level government and these latter forces’ influence on local government. Hence, the thesis avoids some of the critique that has been posed towards proponents of democratic decentralization, such as that it downplays “local social inequalities and power relations”, as well as “view the local in isolation from broader economic and political structures” (Mohan and Stokke 2000:249). Besides, since democratic decentralization is part of global development discourse, implementing such reforms implies that this discourse has an impact on the local level, and the thesis will thus examine this discourse’s influence as well. Cox (1997) argues that the global influence results in “glocalization”, i.e. “simultaneous globalization and localization processes” (Harris et al, 2004:1). The focus on the impact of national and global levels on local government, as well as interaction with existing local power structures derives from the definition of local politics. Local politics is defined as “all politics that are carried out locally, not just in relation to local organs of the state or local governments, but also, for instance, struggles over national and global issues that take place in local contexts” (Stokke and Törnquist 2001). Local politics is a contested game involving several actors at different levels. Consequently, the study of any African democratic decentralization reform should consider local government’s interaction with other levels and existing power structures, and not just emphasize on implementation of policies and formal institutions. Such a study calls for careful contextual analyses, being aware of the impact of informal and formal actors and institutions on local government. By stressing the sway of these actors and institutions in local politics, the thesis is also a critical study of the democratic decentralization discourse presented in previous sections. International donors have learnt from standardized development approaches such as the structural adjustment programs, and they acknowledge that countries’ contexts are significant and adapt their decentralization policies thereafter. Still, national policy- makers, often assisted by international donors do not examine the context sufficiently enough in planning and implementing decentralization reforms. Further, most studies on decentralization in Africa focus on the central state level, and not on rural areas where most Africans live (Bierschenk and Sardan 1998:13, Olowu and Wunsch 8

2003:2). There is thus a need to study the experiences of rural areas in order to understand the complex processes involved in democratic decentralization in Africa.

1.4 An outline of the thesis This thesis will provide an analysis of the result of democratic decentralization in three Malian municipalities in terms of whether such a reform has contributed to substantial democratization or not, and if it has, how and to what extent. I will examine the linkage between democratic decentralization and democratization by using two general analytical frameworks; Törnquist’s framework on substantial democratization and the variables of accountability from Crook’s and Manor’s framework of democratic decentralization, presented in chapter two and five. In chapter two, I only delineate their general explanations that I employ to test the result of the assessment of substantial democratization in chapter seven. Chapter two also briefly introduces the specific theoretical arguments that I use to explain the cases’ outcome. Further, chapter three outlines the research designs used in the analysis, that is, most different and most similar systems analysis. It also deals with their advantages and disadvantages, as well as it demonstrates why Mali is a critical case, and the three sub-cases are critical cases in testing the theoretical arguments. Moreover, this chapter presents the fieldwork, and discusses the methodological challenges of validity, reliability, and generalization of findings. Most of the informants in this thesis are anonymous, due to various reasons. I refer to these with their title/occupation, but the ones that are identifiable by this title are called anonymous. The ones that did not mind me using their name are introduced with their title in footnotes. Chapter four sketches the reform’s most important elements; the processes and actors involved, as well as the actors’ strategies. Turning to chapter five, it describes and operationalizes the two analytical frameworks and the factors by Blair (2000) added to Crook’s and Manor’s framework. The next chapter, then, briefly introduces the background of the three case studies. Chapter seven uses the two analytical framework to analyze these cases, Kontela, Tambaga and Mopti, as to whether substantial democratization has materialized in them and if it has, to what extent. The result of this analysis is that some elements of substantial democratization 9 have materialized in all of the cases, although the quality and scope of these elements vary between the cases. For instance, local government is somewhat accountable in Mopti and Tambaga, but not in Kontela. Besides, Mopti’s local government and state administration has wide scope, but mediocre quality, while Kontela’s and Tambaga’s have low quality and narrow scope. In chapter eight, I discuss the explanations the two analytical frameworks give for why only some elements of substantial democratization have been fulfilled, introduced in chapter two. All these factors are significant explanations for the cases’ outcome, as well as for the differences between the cases. After these general explanatory factors or arguments, I turn to the specific theoretical arguments providing contextual and specific explanations of the cases’ result. The thesis presents these arguments in chapter eight, and assesses them on the findings from the cases in chapter nine. The arguments derive from three theoretical perspectives on African states and politics. The thesis only includes the most relevant arguments of the three theories, related to the problem at stake. One of these theories is Chabal’s and Daloz’s (1999) instrumentalization of disorder, which is to be general for all African states. The theory’s arguments that I use in this thesis are communal belonging, reciprocal and vertical relations, non-separation of public and private spheres, recycling of political elites and some more. Concerning the second theory, Mamdani’s (1996) theory on the bifurcated African state, the thesis assesses two of its central theoretical arguments on the cases; the existence of a rural-urban divide and the lacking balance between decentralization and centralization, participation and representation, and autonomy and alliance. Both these arguments explain why decentralization does not result in democratization in Africa. The third theory is the one by Bierschenk and Sardan (1998) on local politics in Benin, combined with their argument on development brokers (2001). I use their arguments about the existence of several local power arenas, fragmentation of local power and some more. First, I assess all these theories’ arguments on the result from the assessment of the cases in chapter seven. By doing this, I compare the cases, and discuss differences between them that 10 chapter seven present, but also additional ones. Secondly, I assess the explanatory value of the theoretical arguments on the situation within the cases before decentralization, by employing most similar systems analysis. This assessment allows me to depict changes over time in the cases, and measure the ability of the theories to explain such changes. Thirdly, I test the theories’ explanatory value on the entire Mali. Together, these three assessments illuminate the strengths and weaknesses of the theories. Regarding the last chapter of the thesis, chapter ten, it sums up findings and conclusions, and compares the findings from the cases to other democratic decentralization experiences worldwide. Some of these provide additional explanations for the cases’ outcome.

1.5 Conclusion Critique of proponents of democratic decentralization presented in this chapter demonstrate that democratic decentralization reforms tend to ignore the local context’s importance for the result, with its already existing local power structures. This context varies from place to place, creating diverse conditions for decentralization to materialize in democratization. Hence, in order to gain more knowledge on what decentralization has to offer in terms of democratization, it is important to do more research on the local context, which is the rationale of this thesis. The thesis will answer the problem at stake of whether democratic decentralization reform in Mali has contributed to democratization at the local level in the three local governments studied. However, the thesis only examines the lowest governance level, the municipal level to answer this problem. The emphasis is on the local government institution and local actors’ interaction with and influence on local government. Through this focus, the thesis will shed light on the importance of local context and existing power structures for the result of Mali’s reform regarding democratization. 11

2. Frameworks and theoretical explanations This chapter provides a general outline of the analytical frameworks used in the thesis to analyze the linkage between decentralization and democratization in Mali, i.e. whether democratization has materialized from decentralization. The frameworks used are the substantial democratization framework by Törnquist (2002, 2004, see app 2) and the one by Crook and Manor (1998). I have added some elements by Blair (2000) to the latter framework. Only the relevant aspects linked to the problem at stake are included. Chapter five introduces and operationalizes the two analytical frameworks. Besides delineating the general explanations that the frameworks provide, this chapter briefly elaborates on the specific theoretical explanatory arguments used in the thesis. I employ these theoretical arguments, because they offer contextual explanations to the outcome. Chapter eight thoroughly presents these arguments.

2.1 General presentation of assessment frameworks Regarding Törnquist’s substantial democratization framework, it is an actor-oriented approach, because it states that people have to have the capacity to use democratic institutions and rights for democratization to be substantial. Further, institutions and rights are democratic only if they have good quality and a wide scope. This implies that they have to “cover vital issues of public concern” and have to be spread out geographically and on all levels of a state (Törnquist 2004:302). The substantial democratization framework diverges from most mainstream democracy definitions by incorporating the elements of human rights and actors’ capacity, which will be further discussed in chapter five. Since democratic decentralization reforms also aim at increasing citizens’ capacity through popular participation, representation, accountability, and development, the substantial democratization framework is relevant to use in this thesis. If one does not examine the quality and scope of institutions and rights, as well as actors’ capacity, it will not be possible to depict whether a state has reached the aims of democratic decentralization: increased participation, representation, accountability, and 12 development. The substantial democratization framework overlaps somewhat with the other analytical framework used in this thesis, the one of Crook and Manor. I only use their operationalization of the element of accountability. Their framework is employed, because it operationalizes accountability more precisely than Törnquist’s framework does. Chapter five outlines other reasons for using the framework of Crook and Manor.

2.2 General explanatory arguments Both of the analytical frameworks put forward some general explanatory arguments for the linkage between decentralization and democratization. The first and most important factor Crook and Manor (1998) concentrate on is that the national and local, social and political context matter for the outcome. For instance, existing local elites may capture the new local government, and decentralization will thus not lead to equal representation. Other crucial factors for decentralization to materialize in democratization are competitive parties, a widely distributed free press, a professional civil service, and a vibrant civil society. Yet a factor is that local government has a substantial amount of resources. Without such resources, accountability is impossible to achieve, because local government cannot follow up on its decisions. All these factors are important for local government’s accountability, and hence democratization. Concerning Törnquist (2004), his explanations overlap to some extent with the latter ones, but he specifies them, and hence both these two frameworks’ explanations are relevant to use. Törnquist focuses on the linkage between decentralization and democratization by giving four explanations for why democratization does not result from decentralization. The first one is that democratic decentralization reforms often have an elite-based character, which results in unchanged local power relations. This explanation is a critique of the proponents of the “crafting of democracy” argument, which states that democracy results from technocratic engineering of “good” institutions on the foundation of elite-based pacts (Törnquist 2002). The second factor Törnquist outlines is actors’ focus on single issues. The third is actors’ incapacity to link activities in different parts of the political terrain (see app 2). The 13 fourth factor the framework presents is non-existence of public spaces where one can mobilize political issues. Together, these factors emphasize the importance of the local and national social and political context for the result. Further, they contribute to lack of democratization, because if actors only politicize by single interests, and mobilize people in one part of the political terrain no actors will have the capacity to use decentralized institutions in a way that promotes democratization. For instance, if municipal councilors politicize in order to gain resources from being in local government, and mobilize people by clientelism, they leave out openness and equality, and abuse the local government institution instead of promoting it.

2.3 Specific explanatory arguments Having introduced the frameworks’ general arguments on the linkage between decentralization and democratization, I will turn to the specific explanatory arguments on this linkage in Africa. It is important to supplement the general arguments with specific theoretical ones, because the latter provide contextual arguments explaining the linkage. Moreover, the theoretical arguments constitute different approaches to the linkage, and can thus give diverse explanations to the result of the three cases. Two of them, the theories of Chabal and Daloz and Mamdani are general, dealing with all African states and all levels of these states. Bierschenk’s and Sardan’s theory (1998, 2000) is more specific, departing from local level politics in Benin. Still, they argue that general points applying to other African states may be drawn from their perspective (2000:9-10). Before introducing these theories’ arguments, it is however, important to situate them in the theoretical debate to which they belong.

2.3.1 Situating the theoretical arguments in the theoretical debate The historicity and character of the African state (and politics) is the common denominator for the theoretical debate to which the three specific theories belong. The concept patrimonialism, deriving from Weber’s (1997) categorization of states, has dominated this debate. Patrimonialism’s most important trait is “the absence of a distinction between the public and the private domain…in spite of the fact that a 14 structural differentiation between what is public and private exists” (Médard,

1996:80). The patrimonial state is one type of traditional states, where people transfer parental relationships from the domestic arena to the political (ibid.). These relationships result in large-scale patron-client relationship. Patrimonial relationships differ from clientelistic ones, because they involve more people than clientelistic relationships do. Clientelism is defined as “personal dependency based on a reciprocal change of favors between…the patron and the client who control unequal resources” (1976:103). Hence, patrimonialism is a “mode of distribution of public resources in exchange of political support”(ibid.). The three theoretical approaches, as well as other theories provide varying views on patrimonialism. One of the perspectives using the concept, by adding the prefix neo is Médard (1991, 1996) (one of several authors using neo-patrimonialism). In a neo-patrimonial state, traditional patrimonial practices are present in established modern state institutions. The state thus has a dual character. This results in diffusion between political and economic spheres and “personalization of power” (Médard, 1996: 86), where state employees, politicians and other actors with access to state resources use them to serve the patron-client relationships they engage in (ibid.). Such relationships between politicians, state employees, and electorate determine political participation and representation. Without resources, such relationships are not possible to sustain, and the state is likely to dissolve (Braathen et al 2000). Although presenting a general theory on African states, Médard acknowledges that variation between states may exist. Still, his approach is criticized by Amundsen (2001), arguing that the concept of neo-patrimonialism is too general to say anything at all. Bayart (1996:104) agrees with Amundsen, and criticizes all perspectives using patrimonialism for concentrating too much on the colonial state structure for current African states. Instead, he launches his own theoretical approach, pointing out that the post-colonial African state is a combination of pre-colonial state institutions and colonial state institutions. However, the latter have been redefined, giving African states a distinctive character: “the politics of the belly”. The “politics of the belly” is Bayart’s theoretical approach (1996). The belly symbolizes all political actors’ 15 behavior: they “eat” what they gain from political participation. Eating is a visible act, and it locates the invisible; corruption (ibid., Mbembe 2001). Furthermore, government is highly personalized. Consequently, the economic and political spheres of the state are inseparable (ibid.:291). “Eating” is a complex matter, and Africa does not eat uniformly; variations exist between countries (ibid.:325). Badie (2000) criticizes Bayart for his emphasis on the continuance of the colonial state structure in the post-colonial state. Badie outlines a different theoretical perspective, arguing that the effort to import a Western state to Africa failed and the result was a dysfunctional African state (ibid.:131). It was not possible to transfer a European state, because it was based on a particular history and culture of another socio-cultural context (ibid.:10). Instead, African states are different from states on other continents, due to the presence of neo-patrimonial practices, lack of state authority, as well as political participation based on identity and clientelistic strategies (ibid.:168-71).

2.3.2 The specific explanatory arguments of the three theories Turning to the three theories I employ in this thesis, the first to be discussed here is the one by Chabal and Daloz (1999) on instrumentalization of disorder. This theory is one of the latest and most reputed contributions to the theoretical debate briefly outlined in the previous section. The authors argue that democratic decentralization in African states do not result in democratization, because all states are marked by a political disorder where people instrumentalize all their political actions. Instrumentalization of politics is defined as “the profit to be found in the weak institutionalization of political practices” (ibid.:13). The most important traits of this disorder of political instrumentalization is instrumentalization of reciprocal and vertical relations as well as people’s communal belonging (i.e. being part of family, village, etc) (ibid.). The basis of all people’s political participation and representation is the instrumentalization of these factors. Political instrumentalization of these aspects make other traits of the disorder appear, such as instrumentalization of public sphere, resulting in non-separation between public and private spheres (ibid.:14-15), as well as recycling of political elites. Together, these and other factors, described in chapter eight, result in the political instrumentalization of disorder. Turning to 16

Mamdani’s theory (1996), the historical legacy of African local governments, and the colonial period in particular is his departing point. He argues that the bifurcated system of colonial local government has persisted after independence, which explains why decentralization does not result in democratization in Africa (ibid.:218). In this bifurcated state, a decentralized customary despot comprises rural local government. He applies customary law, treating people as subjects. In urban governments, modern law is employed, and people are thus citizens. For democratization to succeed from decentralization, this divide must be dissolved. Further, one needs a balance between decentralization and centralization, participation and representation, and autonomy and alliance for democratization to result from decentralization. Concerning Bierschenk’s and Sardan’s (1998) approach, it argues that the presence of several local power arenas, resulting in fragmentation of local political power and local government lacking institutional and popular accountability, contribute to explain why decentralization does not lead to democratization. Further, as a consequence of fragmentation of power, the state has no legitimacy locally, which constitutes yet an explanation for lack of democratization.

2.4 Conclusion There are many explanations to the linkage between decentralization and democratization. In this chapter, I have outlined the general explanatory arguments as well as the specific theoretical arguments used in this thesis to explain the linkage between decentralization and democratization. The arguments of the two frameworks and the three theories illuminate different conditions for decentralization to result in democratization, as well as why decentralization reforms often fail to facilitate democratization in Africa. Since the thesis aim at a contextual analysis of three Malian local governments, it is important to supplement the general explanatory arguments of the two frameworks with the specific theoretical arguments that derive from the African context. The assessment of these arguments on the cases in chapter nine will show whether they have valid explanations for the three cases regarding democratization. 17

3. Approach and Method The study of Mali’s decentralization reform in this thesis is comparative, because it compares findings from three cases of local governments to answer the problem at stake. The research designs employed, most different systems and most similar systems analysis, are part of comparative method within political science (see f.ex Ragin 1994). I use most different systems analysis to compare similarities and differences between cases, and most similar systems analysis within cases to look at changes over time. This chapter also outlines why Mali is a critical case in Africa, as well as why the three sub-cases are critical in assessing the specific theoretical arguments described in the previous chapter. Further, I discuss the most significant challenges to interpretation of findings: validity, reliability, and generalization.

3.1 The critical case of Mali and the sub-cases There are several reasons for Mali being a critical case in studying decentralization in Africa. First, it is among the poorest and least developed African countries (URL:www.devdata.org). Thus, if Mali’s reform leads to increased participation, accountability, equal representation, and local development, it is likely to do so in wealthier countries. Secondly, the reform’s recent origin compared to other African reforms constitutes a cause for choosing it, because the challenges of implementation are still present, although local government has been established. Thirdly, the importance of the local political and social context for the result of decentralization reforms makes Mali a fruitful choice. If one examines a reform implemented a long time ago, local politics is likely to have changed, and the impact of decentralization would be hard to study. Likewise, if studying a reform where local government has been recently implemented, local government is not prone to have had an effect on local politics at all. By examining Mali’s reform, where local government has existed for four years, I avoid these traps. Moreover, another cause for Mali being a critical case in Africa, and in particular compared to other West African countries, is its stability and persistent democratic efforts since 1992 (Fomunyoh 2001;Tavares 2004). Moreover, a research-based argument, which also 18 justifies the choice of sub-cases, is that it is important to do research on understudied issues and areas as Mali, in order to gain more knowledge on them. Furthermore, concerning the three sub-cases of local governments, these cases are critical in assessing the explanatory value of the three theories. One reason for their critical nature relates to the theory of Chabal and Daloz (1999). It argues that instrumentalization of communal belonging and reciprocal and vertical relations, as well as other features are general traits of African politics. These arguments may be assessed by studying different local governments. For instance, to test the generality of communal belonging, in the form of ethnic belonging, it is thus important to study municipalities that have different ethnic groups to see whether ethnic belonging is equally strong all places. Moreover, considering that family and kin belonging, also included in communal belonging, is often not as significant in urban areas as in rural ones makes it important to study both areas in order to measure whether belonging is equally imperative there. As to Mamdani’s theory (1996), the cases are critical in assessing its argument of the rural-urban divide in local government. Further, because of the theory’s emphasis on the colonial period for current African states, it is significant to look at an urban municipality that existed in this period. Mopti did so, along with six other Malian municipalities. Mopti is, however the only one where researchers have done two case studies, which constitutes a reason for choosing it over other the other urban municipalities. Concerning the theory of Bierschenk and Sardan, the cases are critical in illuminating the complex and informal character of local politics. In order to assess this theory on the cases, any rural case could be critical in Mali. However, Mopti is a critical case, because this theory has not examined urban cases before, only rural. Consequently, if Mopti fits the theory, it will strengthen the theory’s explanatory value. About the general reasons for choosing the sub-cases to employ a most different and most similar analysis, one cause is that there are no published studies on local government after decentralization in the Kayes region to which the two cases of Tambaga and Kontela belong. In order to develop general knowledge of decentralization’s outcome in Malian rural areas, it is fruitful to 19 examine cases not previously studied. Still, it is, as always in research, important to have some literature to rely on in addition to the interviews, and there are two such studies on local governance and politics from the Kayes region before decentralization (Hopkins 1968; Tag 1994). Due to no available literature on the two rural cases, it is imperative to have a third case where studies have been done on its local government, as is the case with Mopti. Further, when dealing with whether these cases represent general trends of all Malian local governments, it is essential to have Mopti to compare with, as well as other cases, on which literature exist. There are two case studies of recent origin on Mopti (Massing 1996a; Bouju 2000). It is particularly important to have literature on the urban case, because one might expect that it is more difficult to capture all processes at play here compared to rural municipalities. Also, owing to that information reaching out to people is a key for the success of decentralization reforms at the local level, it is important to study cases that differ in accessibility. Such differences are likely to lead to varying situations in information provided to the population, which may again affect local government’s accountability. Kontela is very isolated, has no public transport, or any radio station with programs in people’s mother tongue, from which they can receive information. Tambaga is also isolated, but not to the same extent as Kontela. Neither here is there any local radio station. Mopti, on the other hand is easy accessible. In addition, a factor related to accessibility is that state presence is likely to vary, because, in general, it is difficult to get state employees to live in isolated rural areas. Due to Mopti being a city and easy accessible as well as having been a regional centre since the colonial period, state presence is strong. In the rural cases, the state is almost absent. The divergence in state presence has probably influenced the amount of information received about the reform. Further, it may have contributed to variation in state involvement in the local democratization process. Moreover, another reason for choosing the cases, is that is important to study cases that have different ethnic groups, as well as that they vary on the number of ethnic groups present to see whether this variation has influenced attempts at democratization. It is thus fruitful to have one case where only one ethnic 20 group lives, Kontela, and another with several ethnic groups, Mopti. Tambaga is in a middle position with two ethnic groups and important to study as well. Another ground for choosing the two rural cases is that I had contacts there who introduced me to informants. Informants stated that it would be difficult to conduct interviews without these contacts. Further, because of the difficult accessibility of the rural areas in this region, I chose these cases and not other ones, since I knew people living there who could help me with transport. Yet one reason for choosing the three cases is that they vary in presence of NGOs. In Kontela, there are only two NGOs, while Tambaga has a few more, and Mopti has several. It is relevant to examine whether this disparity influences decentralization’s result, for example regarding mobilization of people and in service delivery.

3.2 Research designs Before outlining the research designs, it is imperative to define what a case study is: “A case study is an empirical enquiry that investigates a contemporary phenomenon within its real-life context, especially when the boundaries between phenomenon and context are not clearly evident” (Yin 1984:13). A case study is one of several research designs for doing comparative research on local governments. The design chosen influences the character of information collected. A statistical analysis obtains data that determine variation on variables, while qualitative case studies present findings through interpretation of interviews, documents, and observation. Case study design is a natural choice in this thesis, since its aim is a contextual study of Malian local governments. Further, since I will compare similarities and differences between three local governments in this thesis, I will use comparative method, which is “systematic analysis of a small number of cases” (Collier 1993:107). Comparative method is one of several methods for comparing cases (Andersen 2000). However, most different and most similar systems analysis are particularly relevant for this thesis, because they make it possible to test theories, which other types of comparative case studies do not (f.ex a-theoretical studies, see Andersen 2000:95-99). Comparative method in general can be 21 quantitative or qualitative studies, as well as have different designs (Sundstøl- Eriksen 2000:66). In this thesis, the designs of most similar and most different have a qualitative character. Still, the way of using them resembles quantitative method, because one employs sampling of variation. In most different systems analysis, sampling means that one maximizes differences in surrounding variables in order to find the one similarity in outcome. Variation is minimized in most similar systems design. By such sampling, both designs come close to statistical control of the variation between cases, but the margin of error is not possible to measure (ibid.:107). Most different systems analysis is chosen for the comparison between the cases, because most similar analysis requires a lot of information about the cases to ensure that the cases are the ones where differences are minimal compared to other possible cases. This is difficult to achieve in Mali. It is however possible to employ most similar analysis within cases, as done in this thesis, because information is easier obtainable within cases, and the variation thus possible to control. Most similar systems design is employed to demonstrate changes over time in the cases. Further, by focusing on differences in the explanatory variables, most different analysis illustrates the variations between cases and explanations for these variations. In this thesis, the explanatory variables are the elements of substantial democratization (see ch 5). The reasons for choosing the cases, previously outlined, constitute causes for why cases are likely to vary on the explanatory variables, but yet have the same outcome. The most different design is thus fruitful to use, because in addition to facilitating comparison between cases, it depicts their variation on the extent to which the elements of democratization has materialized. Moreover, the two designs share many of the same problems; they have several variables and consist of few cases (Ragin 1994).

3.3 Triangulation, sources and fieldwork The thesis is based on triangulation of data, which implies using multiple types of sources(Yin 1984). This enhances data’s reliability (ibid.). The 61 semi-structural 22 interviews I conducted from September to December 2003 in the three local governments studied, as well as in Bafoulabé (the district centre Kontela belongs to), and in Bamako constitute the most important information source from the cases. Most interviews are with one informant at a time, but some are also with two or three persons. The informants are local political actors, such as municipal councilors and employees, political party members, state employees, teachers, NGO workers, people in local associations, as well as ordinary people. In Mopti, the number of interviews is substantially lower than in the other two, due to having more literature on this case. I prepared four interview guides in French before going on fieldwork; one for researchers, one for local politicians, one for ordinary people and one for NGO workers and state employees. These guides contained main questions, adjusted in the field by contextualizing questions by adding examples. In Kontela, I used a Norwegian missionary living there as an interpreter in about half of the interviews. She was more neutral than other possible interpreters, because they were all local political actors. In Tambaga, one of the village chief’s sons in Tambaga village was an interpreter in two interviews. He was a municipal councilor, but there were no other people to ask who not political actors were. In addition to interviews, I collected public documents, and the little official statistics available.

3.4 Methodological challenges In this section, I will very briefly discuss the three most significant methodological challenges for the interpretation and use of the findings from fieldwork: Validity, reliability, and generalization of findings. Starting with validity, there are several types, but in general, validity may be defined as testing the accordance between operationalization of a concept and its actual content (Hellevik 1991:159). According to Adcock and Collier (2001) the three relevant forms of validity to consider for political science research are content validity, convergent/ discriminant and construct validity (ibid.:538). I will briefly discuss the validity of the key concept of this thesis: democratization. Content validity is to detect whether “a given indicator adequately captures the full content of a systematized concept” (ibid.). In the thesis, it is relevant to depict whether operationalization of key concepts in the assessment of 23 democratization have this validity. The operationalization of the concepts used is based on other researchers’ testing of various ways to operationalize these. Hence, there is reason to argue that the content validity of the concepts employed is high. Concerning convergent/discriminant validity, it is to detect whether indicators used to assess one concept converge, and at the same time diverge from indicators which assess other concepts (ibid.:540). The concepts used in the thesis are well-known and difficult to confuse with other ones. This form of validity is thus satisfactory. As to construct validity, it “assesses the performance of indicators in relation to causal hypotheses” (ibid.:542). For instance, if democratization has a satisfactory construct validity it implies that the elements of substantial democratization together produce a causal association leading to democratization if they are all present. They do so in the framework. Regarding the challenge reliability, it means assessing whether the instruments applied to measure variables are of good quality (Hellevik 1991:159). If reliability is present, a compatible result is to be achieved by following the same procedure later on or by another researcher simultaneously. The method utilized thus has to be well documented. Further, one enhances reliability if several arguments go in the same direction, thus obtaining consistency (Rubin and Rubin 1995:87-91). This is important in qualitative methods, since it is not possible to assess the margin of error, which elucidates the inconsistency, as in quantitative methods. In this thesis, one challenge of reliability is that a few times I use information from one informant only, because otherwise, relevant information would be lost. However, before employing such information, I check its reliability through relating it to other similar statements in the cases or in literature. Another challenge to data’s reliability is that in interviews in which I have utilized an interpreter, problems might have appeared in the accuracy of information obtained, translating from one language to another. Turning to the third challenge, generalization of findings, it is two-fold. First, it is a question of whether the cases’ findings are representative for all of Mali, or that the rural cases are representative for all other rural cases, and likewise for the urban case. Secondly, it is a question of whether the theoretical arguments tested explain the cases’ outcome 24 as to democratization, which chapter nine discusses. As to the first question, it is even more defying to generalize in this thesis than in other case studies, because of the recent origin of the decentralization reform. Further, generalization of empirical findings is part of a voluminous debate on usage of qualitative vs. quantitative methods. Proponents of quantitative method argue that it is not possible to generalize from case studies because of the limited number of people studied, and because one cannot depict the margin of error (Hellevik 1991:81). Case study supporters have diverse views on generalization. The only one I present here, views case studies as quasi experiments, where the sampling method is the analytical variation between cases (Yin 1984). All in all, even if little research has been done on democratization in Mali, it is possible to sketch some general traits in this thesis due to literature and information from informants.

3.5 Conclusion This thesis employs most similar and most different research designs, aiming at comparisons across and within three cases. I have chosen these designs, because they are fruitful methodological tools to undertake such comparisons. Mali is a critical case, because it is one of the poorest African countries, yet democratically stable. The sub-cases are also critical in testing the three specific theories. Moreover, the sub- cases are chosen for general reasons, outlined in the chapter. The most important methodological challenges of qualitative case studies in general are validity, reliability, and generalization of findings. The latter is more defying in such studies than in statistical analysis, because the margin of error may not be found there. Despite this, it is fruitful to study the three cases in order to outline the importance of contextual features for the result of decentralization reforms, which would not be possible with statistical analysis. This is the strength of the case study design. 25

4. The Decentralization Reform in Mali

The decentralization reform in Mali has two aims: democratization and development. The emphasis on democratization came during the transition period from Moussa Traore’s authoritarian rule ending in 1991 to the establishment of the democratic republic in 1992. The focus on development was a reflex of influence from international donor discourse on decentralization, as well as a response to Mali’s severe economic situation. It was also owing to national politicians admitting the Malian state’s incapacity to provide its citizens with basic needs (MDRI 1998:39). In order to examine the reform’s two aims at the local level, it is necessary to outline the vital factors of the national and local reform process. These factors are the ones needed as a background for the analysis of the cases in chapter seven and the assessment of theories in chapter nine. I will delineate these factors by asking specific questions, which the first section of this chapter presents. The remaining sections will then answer these questions.

4.1 Questions on the decentralization reform The questions in this section are the ones I will answer in this chapter. I have chosen these, because they illuminate important aspects of the national and local level processes of the reform.

-What attempts have earlier regimes made at decentralization? What were the already established institutions at the local level before the current decentralization reform?

-Who were the actors in the initiation of the reform process and what were their strategies?

-How was the reform organized? What was the role of the national state administration? What was the opinion of state employees on the reform?

-How is local government structured? What is the responsibility of local state administration, village chiefs, and local government? What is the judiciary in the 26 municipalities? What is the municipalities’ revenue base? -What are the challenges to the reform?

4.2 The initiative to decentralize - actors and strategies 4.2.1 The participants at the National Conference The 1991 regime change that was brought forward by uprisings of urban civil society from 1990 and on led to the third decentralization initiative after independence in Mali. The previous attempts at decentralization (deconcentration) came in the First (1960-1968) and Second republics (1968-1991), but both efforts failed, because they did not manage to change the centralized state character (Rawson 2000:265). The Malian state had since colonial times been a centralized Jacobin state, based on the French model (ibid.). The repression from colonial times persisted at the local level after independence, through the state system with single party rule, described in chapter nine. Taken into consideration this history, the third initiative of decentralization focused on democratization. This initiative came about in the National Conference organized in 1991 after Traore’s overthrow. This conference had participants from civil society, political parties and the military. It drew up a new constitution and a governance system that recognized the fundamental principles of decentralization. The population approved the constitution by referendum in 1992 (Sy 1998:11). The national conference recommended that decentralization should be a part of the national democratization process (Rawson, 2000). This implied that the reform was to be one of democratic decentralization, which it actually is. The civil society associations that participated in the national conference constituted a force from below in the reform. Further, the claim of autonomy from the Tuaregs in Northern Mali was another force from below, and decentralization was a strategy to deal with the Tuaregs’ demand (Seely 2001; Vedeld 2003:193).

4.2.2 National government and political parties The transitional government established after the national conference saw decentralization as a strategy of compensating for the repression by earlier regimes, 27 as well as for regaining state legitimacy in the population. This was also the view of the first national government, installed after the 1992 elections (MDD 1997). The “new political context” (Rawson 2000:265) that the national government represented was decentralization’s “driving force”(ibid.). Most political parties recognized the need to decentralize, although some opposition parties opposed the reform, claiming that it was not well implemented and would therefore not work (ibid.:280). There were about 90 political parties in 2003 (Kassibo 24.09.03 [interview]), but only a few serious ones (Vengroff 1993:544). Scissions in existing parties have created most parties (Kassibo 24.09.03 [interview]), and the party system is fragmented (Sandbrook 2000:34). Splits and internal conflicts in parties have made cooperation in local as well as national politics difficult, and have slowed down the decentralization process (Kassibo 24.09.03 [interview]).

4.2.3 The role of the state administration State employees at the central and the local level were significant actors in planning and implementing decentralization. In order to facilitate a dynamic reform process, an independent ad-hoc structure was set up in 1993, the Decentralization Mission (MDD) (Kassibo 1997:5). The resistance of central and local state employees towards the reform was widespread, because to them, decentralization meant loss of privileges and power (Rawson 2000). In 1994, the government had to relocate MDD and place it under the President to ensure independency in its work (Lippman and Lewis 1998). MDD’s strategy was to involve people, trying to have a bottom-up approach to the reform (Sy 1998:16). However, local state officials’ resistance towards the reform hindered their work, which is one of the causes for devolution of services moving slowly (UNDP 2003:12).

4.3 Elements and processes of the reform 4.3.1 MDD and legislation MDD followed up on the National Conference’s work on decentralization by formulating the legal components of the reform, the laws, which the National Assembly passed. In 1993, the law establishing the new governance levels; the 28 regions, the districts and the urban and rural municipalities was implemented. The municipalities replaced the arrondissements, and the act recognized devolution of power to these levels. They were to have “corporate personality and financial autonomy” (République du Mali, 1997: 13). Still, it was not until 1996 that the 682 new municipalities were established. Together with the existing 19 urban municipalities, Mali had 703 municipalities in 2003; 666 rural and 37 urban ones. Furthermore, the international donors had an important role in the decentralization process through their assistance to the MDD and the state administration.

4.3.2 Participation by the population The reform’s most inventive trait was the redrawing of territorial boundaries, which implied regrouping villages into municipalities. This gave the reform an approach from below. The process was however, organized from above, by the MDD, who set up the criteria of cultural, economic and geographical viability to be followed. MDD also decided that each arrondissement was to have a local commission that arranged the redrawing (Hetland 2001:87). The commission and the mobilization and information groups established at all levels were the participatory elements of the reform. The commissions and the groups were to be representative of the population and had to include the area’s state representative, at least two civil society representatives, five additional people resources, as village chiefs, and technical experts, as well as women (Diallo 1995:27). Many places people participated in the groups and commissions and the meetings they held. However, participation by women and youth was lower than for men (Bako-Afari 1998; Béridogo 1997). In some instances, local political elites dominated commissions and groups, and were preoccupied with benefits they could gain from participating, such as power (Kassibo 1997; de Langen 2003; Massing 2000). Hence, for some of the people in these groups and commissions, notably local political elites decentralization was a strategy to recapture the power they had enjoyed under colonial rule (Rawson 2000:269). These elites did not bother to inform the population (Béridogo 1997:28; Sy 1998:17). There was also deficiency in information from the state some places, and the information that came was not always in their mother tongue, making it 29 impossible to understand (Kassibo 1997). For several people, decentralization was thus a “return of the power to the people” back to pre-colonial self-government (Kassibo 1997:267). Further, the redrawing involved a final break away from the colonial period, by municipalities changing their names and boundaries, originating from this time. Chapters seven and nine will elaborate on the variety of local processes. In some cases, local commissions did not abide to the criteria of the redrawing. This was advantageous, because it made the municipalities socially coherent, while in others it implied that economic viability was far from ensured (Koné 1997; Kassibo 1997). The redrawing also revived old local power structures, which might have destroyed the potential for democratization locally (Béridogo 1997:30; Rawson 2000). Also, the redrawing created about 200 conflicts within and between municipalities, and 50 of these resulted in proposals to redraw municipalities that the National Assembly are to deal with (UNDP 2003:18; Ouane1 21.11.03 [interview]).

4.3.3 Establishing new institutions and devolving power The decentralization reform reorganized the entire territorial structure of the country, dividing Mali into 8 regions, 49 districts and 703 municipalities. Each region is composed of districts, while the districts constitute of several municipalities, the lowest formal level of government. The role of the state at the various levels is one of tutelage, supervising the devolved authorities and controlling that their decisions abide by the law (République du Mali 1997:55-59). Rural municipalities constitute of villages, while urban ones of wards and pastoralist municipalities are composed of fractions. These villages/wards/fractions are not a separate administrative level, but an important entity for consultancy and conflict resolution, over which the village/ward/fraction chief and his council presides. The regions, districts and municipalities are responsible for economic, social, and cultural development, including, environmental protection and infrastructure development, and many other

1 Coordinator of the World Bank financed project of urban development and decentralization. 30 tasks. Until 2004, the state has transferred the services of education, health and water to the municipalities. Within these services, the municipalities only took care of budgeting, planning and making contracts with teachers and health employees in late 2004. Moreover, the population elects the municipal council every five years in proportional elections of one tour. The council elects the mayor, and he/she is the municipality’s executive leader and head of the municipal administration. The mayor leads council meetings, and cooperates with the local state representative. The mayor is also responsible for police and judiciary at this level. He has three assistants helping him in his work, elected by and among the municipal councilors. All municipalities also have a general secretary who is the mayor’s most important advisor, together with the two accountants: one for expenses and one for revenue. The municipal council is to organize its work on commissions. All councilors are to participate in one of these. Devolution in terms of transfer of personnel and resources is to be “the essential translation of decentralization” (Sy 1998:22), giving local governments a decisive role in development of their municipality. The inherent problem is deficiency of qualified personnel and financial resources even at the national level. Thus, there are few resources and personnel to transfer. This impedes on local government’s accountability and development, because without competent administrative staff and resources, local government cannot carry out their decisions. As to the village/ward/fraction chiefs’ role, the local government act integrates these chiefs, because it states that the municipal council is to consult this chief on several matters, as transport, natural resources, and development (République du Mali 1997:17). The chiefs are representatives of the state in the village/ward/fraction and assist in tax collection.

4.3.4 State assistance and local revenue base Development was one of decentralization’s two most significant aims. Two institutions were set up to facilitate development: Centres Conseils Communaux (CCC) and Agence d’Investissement des Collectivités Territoriales (ANICT). CCC’s tasks are to inform the municipalities of their responsibilities and help them apply for 31 support from ANICT (République du Mali 2001). ANICT is a financial organ to which the municipalities apply for funding of small-scale infrastructure projects, as schools and health centers. The funds are an important contribution to development in the municipalities (Vedeld 2003), because they have few other resources to rely on. Although the municipalities have been given financial autonomy, implying that they are free to collect additional taxes and fees on local markets, these do not amount to much and most municipalities rely on state transfers (UNDP 2003:122). Still, state transfers do not amount to much, because the population refuses to pay taxes (ibid.). The most important reason for the refusal is the misunderstanding the political party ADEMA created when it abolished the head tax in 1992. In Malian languages there is only one word for taxes, and people thought that all taxes had been abolished and stopped paying them (ibid.:123).When the government discovered the misunderstanding, information campaigns were run, but they did not reach all people (ibid.).

4.4 Challenges to the reform-contextual factors There are three important challenges discussed here, the social structure and existing power dynamics, management of natural resources and property, and electoral participation. As to the first challenge, the local context has had an impact on Mali’s social structure and existing power dynamics, because local socio-political processes have formed relations between different social groups and determined power structures for centuries. The caste system is present most places, although formally abolished. Its composition differs between ethnicities and places, as well as its usage in politics, seen in this thesis. Endogamy and occupation define to which caste one belongs (Béridogo 2002:9; N’Diaye 1995:38). It is thus a system of inequality, and contrary to the idea of democracy (ibid.:7). All ethnic groups have a noble caste (often called founders), at least two castes of artisans (blacksmiths and shoemakers) and the caste of ex-slaves. Exchange of gifts and services characterizes relationships between the upper caste of nobles and the other, lower castes. Lower castes give gifts and mediate in conflicts. In return, the nobles protect them (Béridogo 2002:18). Nobles usually possess the administrative, political and religious powers. They are 32 the ones the village/ward/fraction council elects as chiefs and nobles usually serve on this council (CCC Bafoulabé 2001:9). The chief is generally a descendent of the founder family of the village/ward/fraction, and the principles of lineage and age decides whom the village/ward/fraction council elects from this family. Consensus characterizes decision-making on this council, which meets frequently (Lippman and Lewis 1998). In many places, the system hinders all other castes, as well as “foreigners” from participation on council (Beridogo 1997:16-17). The latter group constitutes of people of ethnic groups that live in an area where another ethnic group dominates (ibid.). Social mobility is rare in Mali, because descent, gender and age are important for a person’s status (ibid.). Still, the caste system has “adapted to social evolution” in urban areas, turning into clientelistic relations with money and property as essential issues (ibid.:23). In rural areas, the caste system has not changed much, and does not involve money to the same extent as in urban ones (ibid.). Nonetheless, in all areas, it plays a role in politics, because a vote-catching relationship between lower and superior castes exist in national and local politics (ibid.: 24). Regarding the second challenge of the reform, management of natural resources and property distribution this is an important issue in Mali, because natural resources and property are the most valuable assets for Malians. In most places, the village/ward/fraction chief distributes property and natural resources, although these formally belong to the state, unless people have acquired usufruct over them (Vedeld 2002). Customary law constitutes the foundation for this management (Ribot 2002). Customary law is an oral and fluid system of laws that have been established throughout history and varies between localities (ibid). The local government act recognizes customary law as a mechanism for solving civil disputes, either by village/ward/fraction chiefs and their councils, or by the mayor. The flexible practice of law allows for a dynamic approach to conflict resolution. Decentralization is to transfer responsibility over property and natural resources to municipalities (République du Mali 1997).With the exception of some urban municipalities it has no taken place yet, because conflicts are likely to appear between village/ward/fraction chiefs and municipal councils (Massing 2000; de Langen 2003; Kassibo 2002). 33

As to the third challenge of electoral participation, the low participation in municipal elections demonstrates that this challenge is a major one. Mali held its first multiparty municipal elections in the 19 existing urban municipalities in 1992. Only 37% voted (Bértrand 1992:21). In 1998, elections in these urban municipalities took place again, and 38,8% voted (Bértrand 1999a;Hanke 2001:198). These elections were followed by elections in rural municipalities in 1999. The main opposition coalition, COPPO boycotted these elections, protesting against fraud in the 1997 national elections (Hanke 2001:195). In the last municipal elections in May 2004, 43, 8% participated, being the highest rate in ten years (Jeune Afrique 2004).

4.5 Conclusion Although there were elements of initiative from below in Mali’s decentralization reform, the reform was largely a top-down process, because of the dominant role by the state administration and MDD in its planning and implementation. There are and have been obstacles to the reform. At the national level, instability of political parties, which slows down decision-making in the national assembly, resistance by state employees, as well as lack of resources constitute hindrances to the reform. At the local level, the conflicts the redrawing has generated, lack of information about the reform, and elite capture of local government are important problems. In general, the crucial challenges in 2004 are devolution and deconcentration of responsibilities, interaction between the new local governments and already existing power structures, equal participation and representation in local government, as well as fair distribution of property and natural resources. As to the latter three challenges, they require restructuring of local power relations and establishing another institutional organization of local government. For deconcentration and devolution to work, local governments have to get more resources than they have in 2004. According to informants, national government has not dealt effectively with these challenges, which is a serious weakness in their work on the reform. This weakness is also an explanation to the result of the three cases regarding democratization, as the remaining chapters of the thesis will demonstrate. 34

5. Operationalization of analytical frameworks

This chapter provides a detailed outline of the two analytical frameworks presented in chapter two. These frameworks are used to assess whether decentralization has contributed to democratization in the three cases, and the extent of its materialization. The two frameworks complement each other. The operationalization of the frameworks in this chapter clarifies how I will employ them in the assessment of the three cases in chapter seven. The strength of both frameworks is that they allow for measuring the influence of informal actors and institutions on local politics. It is not possible to discuss Mali’s local politics without looking at these.

5.1 Assessing and operationalizing substantial democratization Törnquist’s (2002, 2003) substantial democratization definition and framework focuses on institutions, rights, and actors’ capacity to use these. The emphasis on rights builds on Beetham’s (1999) definition and assessment of democracy, which includes civil and political rights. Törnquist’s framework assesses 36 factors dealing with institutions and rights, as well as actors’ capacity (Törnquist 2003). This thesis only includes the most important factors of the institutions and rights with which the framework deals. The factors I chose to focus on are the most vital aspects of democratization as seen by several democracy definitions. Concerning rights, Törnquist includes all human rights, but it is too encompassing to assess all human rights embraced by the framework. Consequently, the thesis only assesses five crucial civil and political rights, described in the next section. As to institutions, the thesis measures the existence, quality, and scope of local government, state administration and civil society in the three cases. This is done by studying whether government and state are open, accountable and representative, whether municipal elections have been free and fair, and whether civil society engages the entire population and is able to influence local government. To assess institutions’ accountability, I have used Crook’s and Manor’s framework instead of Törnquist’s (see sect 5.2) Furthermore, I assess the existence, quality and scope of judiciary, rule of law and citizenship. 35

Measuring the quality of institutions and rights implies studying their performance, while assessing their scope means examining how widely they are spread out geographically in the municipalities, as well as if they capture vital public issues (ibid.). Even though Törnquist’s framework looks at how widely they are spread out on all levels, I will only study the municipal level, since the thesis concentrates on this level. In addition to the assessment of institutions and rights, Törnquist adds political actor’s capacity to use institutions and rights to further democratic or instrumental aims as an element, which separates his framework from Beetham’s. To assess actors’ capacity is imperative in order to discuss whether a municipality is in a state of substantial democratization, because democratization is not substantial if people are not willing and able to use institutions and rights. Four factors assess political actors’ capacity: 1) actors’ capacity to “be present and link activities in different parts of the political terrain”; (Demos 2004:5; see app 2) 2) actors’ politicization of issues and interests; 3) actors’capacity to “mobilise support for policies with regard to such interests, issues and values”; (ibid.:5) 4) how political actors relate to and use institutions and rights (ibid.). Concerning the first factor, the political terrain comprises the central and the local level where actors may be present. Each level has three sections: the self- government, the government and the business section (Törnquist 2002; see app. 2). In between is the public sphere with the civil and political societies. In the political ones, actors such as political parties try to influence state and government directly. However, political parties and other actors in political societies, for instance social movement-based political parties, may also link activities in political and civil societies. Influence of civil societies on state and government is indirect, according to the framework (Törnquist 2002:37). These societies constitute of various types of organizations, such as social movements and primordial associations, but excluding political parties. Turning to the second factor measuring actors’ capacity; politicization of issues and interests, there are three types of issues and interests actors politicize: single issues/specific interests, ideology/collective interests, and values/communal loyalties (Törnquist 2002:41). Actors may politicize all three types at the same time. As to the third factor, actors’ mobilization of people, they mobilize 36 in three ways: by incorporation, integration, or alternative patronage. Incorporation means that elites in parties, among others, integrate people into these. Integration implies that “relatively autonomous broad popular movements” (Törnquist, 2002:39) mobilize people. Alternative patronage (strong incorporation) signifies incorporation where integration is the aim (ibid.). Out of the three possible politicization strategies actors employ and the three potential mobilization strategies, three possible joint strategies of politicizing issues and mobilizing people materialize on a continuum (ibid.:37; see app.2). The joint strategy of actors’ politicization of single issues, collective ones or communal loyalties and actors’ mobilization of people by incorporation ends up in a joint strategy lying on a continuum in between clientelism or populism. Whether the joint strategy tends towards clientelism or populism depends on how actors politicize issues and mobilize people. Clientelism involves a mutual relationship between two parties, where one of the parts are either a patron or a boss, “delivering protection in return for votes” (ibid.:39). By including bosses, Törnquist extends the definition of clientelism (see sect. 2.3.1). Nevertheless, he emphasizes that bosses and patrons differ from each other on the nature of power they enjoy. A boss has monopoly of power in his district, is more coercive and dependent on political power than a patron, who is one among several in a district and governs by socio-economic resources (f.ex Sidel 1999). Moreover, the second joint strategy of mobilization and politicization is on the continuum between incorporation and integration. Single issues, collective interests or communal loyalties of politicization and mobilization by alternative patronage comprise the foundation of this joint strategy. In the third joint strategy, the one of integration, actors mobilize people and politicize single issues, collective interests or communal loyalties on a continuum between networks and organized integration. As to the fourth factor assessing actors’ capacity; actors’ relation to institutions and rights, there are four ways of relating to these: use, abuse, promote or bypass them. If actors use democratic rights and institutions to promote democracy as well as their instrumental aims, they are pro-democrats and users. If they abuse them, they are anti-democrats. They are consumers/users if they draw on 37 institutions and rights without promoting democracy (Törnquist 2002:32). If they do not use institutions and rights at all, but find other ways to promote their interests or mobilize people, they bypass them (ibid.). In addition to all the four factors assessing actors’ capacity presented in this section, there are three conditioning elements for actors’ capacity that are not intrinsic to democracy as the four factors mentioned above. These three additional elements are actors’ opportunity structure, their sources of power, and their capacity to transform power into legitimacy (Törnquist 2003). Stokke (2002) has explored on these factors, and I will use his point of departure for assessing them. The first element, opportunity structure, is “the opportunities and hindrances” actors meet in political systems, “such as the degree and openness of the system, the presence of allies and the risk of repression” (Törnquist 2003:7). The second element, the actors’ sources of power includes two aspects. The first is actors’ “habitus”. Habitus originates from Bourdieu (1977) and is “a system of internalized social norms” that “incline actors to act in certain ways” (Stokke 2002:5-6).The second aspect of power sources is the capital actors possess. Actors may have three forms of capital; social, economic, and cultural capital. Social capital is “social resources in the form of networks or contacts based on mutual recognition”. Economic capital is “material wealth in the form of property, money, and shares (etc),” while cultural capital is “informational assets in the form of knowledge and skills acquired through socialization and education”. The third element Törnquist’s framework sketches as conditioning actors’ capacity is the transformation of the three forms of capital into symbolic capital. Symbolic capital is defined as “legitimate authority in the form of prestige, honor, reputation and fame” (Stokke 2002:8-9). An actor obtains symbolic capital if he/she is able to transform economic, social or cultural capital into legitimacy (Bourdieu 1977). Having explored how this thesis will measure and operationalize actors’ capacity it is important to define and operationalize the institutions and rights of Törnquist’s framework. Civil and political rights is one of the key terms that I use from Törnquist’s framework to assess substantial democratization in the three Malian cases. These rights ensure political and civil being in a society. Following Törnquist (2003), this thesis will only measure the 38 existence, quality and scope of five such rights: equal treatment before the law, freedom of trade unions, individual freedom from physical violence and fear of it, freedom of religion, culture and language, and freedom of speech, assembly and organization. Concerning the institutions, judiciary and rule of law imply that there is “subordination of government and public officials to the rule of law; equal and secure access to justice and independent courts and judiciary” (ibid.:4). The district court is Mali’s lowest formal level of judiciary. At the municipal level, the village chief and the mayor together with his councilors mediate in conflicts. This thesis examines these two instances and the rule of law in them, since the thesis’ emphasis is on the municipal level. By assessing the judiciary and rule of law, I also assess the existence, quality and scope of citizenship, because in a democratic regime citizenship means rule of law, i. e. equal treatment of all citizens before the law. As to another institution of substantial democratization, free and fair elections, they comprise ”regular, competitive, open and just registration of voters’ candidates, parties, and voting procedures, as well as competitor’s fair chances to communicate and participate” (ibid.:114). Regarding civil society, one may define it in many ways (Tostensen et al 2001). In this thesis, I have included only the most relevant factors from Törnquist’s assessment of civil society. According to Törnquist, civil society includes all associations, trade unions, and social movements in a municipality or country. Assessing civil society’s existence, quality and scope, i.e. whether it is democratically oriented or not, implies examining whether civil society is present in the cases. If it is, I measure its quality and scope by two factors: people’s participation in civil society and its influence on local government. As to the aspects of open and representative government, this is assessed by studying whether it is open to all social groups and genders and whether there is equal representation of these groups. Assessing whether the administration is open and representative administrative means looking at whether its work is transparent and reflects people’s needs and demands. 39

5.2 Assessing and operationalizing democratic decentralization As previously mentioned, I will employ elements from Crook and Manor framework (1998) in the assessment of whether decentralization has contributed to realize substantial democratization in the three cases. This framework has set the agenda for assessing results of democratic decentralization reforms. The aspects I use from this framework are participation and institutional performance, because together these measure the accountability of state and local government in the cases (ibid.10). They assess political accountability, i.e. from municipal council to the electorate and public accountability, i.e. from municipal administration to the electorate. I use the framework to examine accountability because the operationalization and assessment this framework has of accountability is more precise and contextualized than the one in Törnquist’s framework. It is more contextualized, because it derives from studies of West African and Asian democratic decentralization reforms. Further, its strength compared to other general frameworks on democratic decentralization is due to combining quantitative and qualitative research methodology in assessing democratic decentralization. Moreover, the framework is useful, because its variables focus on local level processes, considering central level’s impact as well. Besides, its variables include informal institutions and actors, which are important since the reform studied is of recent origin, and informal institutions and actors are thus likely to influence local government. This assessment from below is unique, because other assessments of decentralization tend to use central level as a starting point. Furthermore, the framework is relevant to employ, because it has a comparative character, just as this thesis. It analyzes democratic decentralization through four concepts: participation, representation, institutional responsibility and institutional performance. Then, several qualitative and quantitative variables measure these concepts (ibid.:10). As previously stated, this thesis only employs the concepts of participation and institutional performance, since it uses the framework to assess the accountability of the state administration and local government. Only the 40 qualitative variables that operationalize and assess the concepts of participation and institutional performance are included in this thesis, because the statistic material required to examine the quantitative variables of these concepts does not exist in the cases. Further, some of the qualitative variables used have been simplified or combined, because otherwise, they would be too extensive to employ. Crook and Manor have five aspects for measuring participation, and these aspects have one or more variables each. Altogether, I use eight of their variables assessing participation (ibid.:16-18). The first variable of participation is electoral participation, and here I concentrate on participation in municipal elections, which the factor of free and fair elections in the substantial democratization framework includes (ibid.:16). The second variable is participation in informal and formal meetings on the village council or in other village meetings (ibid.). The third variable is direct participation in activities related to the municipality, such as contact with municipal councilors, the mayor, or municipal employees (ibid.:17). The fourth variable is indirect participation, which means participation in associations (ibid.). The civil society factor of substantial democratization incorporates it. The fifth variable examines the contact between municipal councilors and the electorate. The sixth variable measures political participation of marginalized groups. In the cases, this includes women, young people, and lower castes (ibid.). The seventh variable focuses on the frequency of contact between the electorate and the councilors. The last variable assesses to what extent the electorate know the municipal council and councilors. To the latter, I have added the aspect of information from the council to the electorate by Blair (2000). Blair analyzes this by studying whether local government has informed the population about its decisions through public meetings, posters, radio, or newspapers. It is imperative to look at this aspect in order to achieve a comprehensive evaluation of accountability. As to the concept of institutional performance, which constitutes the second concept in determining local government’s accountability there are three dimensions that examine it: output efficiency, responsiveness, and process (ibid.:18). Output efficiency has two variables, but I will use only one of them in this thesis, because the other is too 41 extensive and I lack information on it. The variable I utilize compares local government’s work to the achievements of the previous local state administration (ibid.).Moreover, the dimension of responsiveness has four variables, but they require an extensive survey, which would be difficult to conduct on fieldwork. Thus, responsiveness is measured by two variables. The first variable of responsiveness is to evaluate the local government’s work by interviews. The second is to measure people’s opinion about the local government’s work, also based on interviews. The third dimension assessing institutional performance is process, i.e. the way government and administration rules (ibid.:19). This variable concentrates on corruption and the use of law, among other aspects, treated under the element of judiciary and civil and political rights in the substantial democratization framework. In addition to participation and institutional performance, I add political parties as a variable of accountability, drawing on Blair (2000), because it is important to study whether they may enforce local government’s accountability.

5.3 Conclusion Since the aim of this thesis is to examine whether Mali’s democratic decentralization reform has contributed to democratization at the local level in three Malian local governments, it has to employ analytical frameworks that analyze this question. There are however many ways to define and assess democratization and decentralization, but few that include the interaction with and influence from informal institutions and actors on local institutions and rights. Since the thesis emphasizes on the importance of informal institutions and actors in local processes at play after implementation of local government, it is important to use frameworks that deal with these institutions and actors. The two frameworks presented in this chapter and used in the analysis in chapter seven are able to analyze the impact of such actors and institutions in various ways. Besides, they supplement each other in assessing these actors, and employing both of them is thus necessary. Together, they provide a good basis for the contextual assessment of whether decentralization has contributed to substantial democratization in the three cases of local government. 42

6. Synopsis of three local governments

This synopsis provides a brief background on the three cases of Kontela, Tambaga and Mopti. The background information presented here on the general character and composition of local government, the redrawing process, and the information provided during the reform, are all significant factors needed for the assessment in the following chapter. They are important elements, because they may have had an impact on the democratization process in the three cases. In chapter seven, I will discuss them in detail. Most of the information in this chapter is based on interviews conducted in the three municipalities.

6.1 The case of Kontela Kontela municipality is situated in a desolated area of the Kayes region. The municipality has 17000 inhabitants and consists of 28 villages and some hamlets (République du Mali). The population belongs to the ethnic group Khassonké (République du Mali 2002:8). Most people are farmers (CCC Bafoulabé 2000:5). The village of Goundara is the administrative center of the municipality. It has one state representative and one technical consultant from the state. The redrawing process added three villages to the municipality. The state representative and the village chiefs of these villages decided on this inclusion in closed meetings, which did not involve the population (state employee 11.10.03 [interview]). When they later announced the decision made, the population of the three villages protested, because they preferred to become part of another neighboring municipality. The resistance towards the decision ended up in a small revolt, as well as people refusing to pay taxes, voting in elections, and expelling all visitors from state or municipality. This example shows the exclusion of Kontela’s population in the redrawing process. It only involved the previous mayor, the state representative, some village chiefs, and a few others. Local information groups were not set up. Most people had therefore never heard about the reform, or about what the municipality was to do, according to several informants. There was absence of information to the population in the entire district to which Kontela belongs (Guindo 1999). As to the municipal elections in 43

1999, there were four political parties in the 1998 municipal elections: ADEMA, PARENA, BDIA, MPR, and US-RDA (see abbr.). ADEMA got five representatives, BDIA three, PARENA one, and US-RDA eight (municipal employee16.10.03 [interview]). After the elections, two of the councilors changed parties. As to the mayor’s election, which takes place in the council, there were two candidates, one from BDIA and one from ADEMA. BDIA allied themselves with US-RDA before these elections, thus having majority. However, a person from ADEMA’s district office offered councilors money to change side, and some did. Hence, the result of the election was 8-8 (one councilor was absent). The next two rounds ended up with the same result. After three rounds of election, the law states that the oldest candidate wins (République du Mali, 1997:19), and ADEMA’s candidate became the mayor. He deceased in 2003, and his first assistant replaced him.

6.2 The case of Tambaga Tambaga municipality is also situated in the Kayes region, in the Kita district, and is less isolated than Kontela. The ethnic group Malinké dominates the area, but also the ethnic groups of Fulanis, Bambara and Diawambé live there (Commune de Tambaga 2001). Before decentralization, Tambaga was one of the four sectors in Kobokofata arrondissement (ibid.). When boundaries were redrawn during the decentralization reform, the local political elite of the villages in this sector decided to unite in one municipality (councilor 28.10.03[interview]). The municipality has 9593 inhabitants and consists of ten villages (Haugsjå 2002:1). The village Tambaga is the center of the municipality. During the decentralization process, the state informed people through meetings held at the village chief’s residence, but only heads of families participated in these (village chief 29.10.03 [interview]). Information to the population was thus scarce. The state did not even inform the municipal councilors of their tasks after the municipal elections (teacher 28.10.03[interview]). The state only has one representative in Tambaga, and this representative has the responsibility for two other neighboring municipalities as well. Moreover, the municipal council has eleven representatives. Initially, it consisted of five representatives from PARENA and six from ADEMA, but some councilors changed to the parties URD and RPM, 44 thus adding two parties to the council. The mayor’s election turned into a conflict between the current mayor (ADEMA) and another candidate, because the latter had stolen several cows from the former (councilor 28.10.03 [interview]. As to the work of local government, the municipal council holds four meetings a year in addition to the budget session at the end of the year (ibid.). The municipal council is to divide its work between commissions, but neither of these function, according to the mayor and the municipal employees. Tambaga shares its state representative with two neighboring municipalities, and its office is in another municipality. The state is only present during tax collection and in some municipal council meetings.

6.3 The case of Mopti Most of Mopti municipality is situated in the Niger river delta, in the Mopti region. Mopti has been an economic, political and cultural center from the 17th century under the Malian Empire (Massing 1996a). The population is composed of several ethnic groups. Still, the population is divided in three social groups that cut across ethnic lines. Rich artisans, founding families, a village chief, and the ward chiefs constitute one group, while small traders and artisans another, and poor people a third. According to Bouju (2000), a mutual clientelistic relationship exists between the first and the two latter groups. The municipality has about 100000 inhabitants (République du Mali 2003:385). There are eleven wards, each with one ward chief. In 1920, the French established the municipality of Mopti, but it was not until 1955 that an elected municipal council and a mayor replaced the administrative mayor and Mopti received some devolved powers (Massing 1996a). In addition to the municipal council, Mopti has a village chief and council, elected by the lineage principle (see ch. 4). The village chief comes from one of Mopti’s founding families. Ward chiefs are part of this customary system, serving as the municipality’s administrative representatives in wards. They do not have to be of the founding families, but people living in the wards have to trust them. However, all current ward chiefs are nobles. The redrawing of boundaries in the current decentralization reform only brought with it discussions of altering boundaries, but no changes were done. Since local government had existed for a long time, people 45 knew about its functions, and were frequently in touch with it through the services provided, such as in the decentralized civil affairs offices in five of the wards (NGO worker 11.11.03 [interview]). Further, the state has a substantial presence in Mopti. Regarding Mopti’s municipal council, it has 29 councilors. The 1998 elections led to a slight alteration in the council’s composition, but the mayor was re-elected. During election campaign, candidates went from door to door to talk with people and organized meetings (councilor 11.11.03 [interview]). ADEMA got eleven representatives, RDT three, PARENA three, RAMAT one, PDP three, and RND eight. Some of the representatives changed parties after the elections, because they had not received any resources from their respective parties, or because they preferred another party (councilor 11.11.03 [interview]).The municipal administration has about a hundred employees, and provides numerous services, such as in health and education. Although the municipality manages property, the village chief, ward chiefs and founding families are involved in sales and distribution of property, (Bouju, 2000) which later chapters will describe.

6.4 Conclusion This synopsis has demonstrated that there are great differences between the two rural cases on one hand and the urban one on the other concerning the redrawing process, information provided on the reform, and the work of the three local governments. The difference in experience of local government and state presence are two of the reasons for the variation between the cases. These features make Mopti in an overall better situation to handle the challenges of decentralization than the other two municipalities. However, there are similarities between the cases as well, such as that their redrawing process was top-down, because the population was not consulted. This neglect and the variation between the cases are likely to have affected the cases’ outcome with regard to democratization, which the next chapter will show. 46

7. Assessment of substantial democratization in three local governments In this chapter, I will analyze three cases of local government (municipal council and administration) in the municipalities of Kontela, Tambaga and Mopti in Mali. My assessment employs the two analytical frameworks presented in chapter five, and specific questions outlined from these follow below. First, based on Törnquist (2002, 2003, and 2004), I assess the existence, quality and scope of institutions and rights, regarding the aspects of representation, accountability, and openness. However, concerning accountability, Crook and Manor (1998) have a more precise operationalization of it that the thesis will use. One separate section treats accountability, by using the variables of the concepts participation and institutional performance. I have added the aspects of information and political parties described by Blair (2000) to the variables of participation, because these are relevant factors that Crook’s and Manors’ variables do not capture. Secondly, in this chapter, I will measure political actors’ capacity, outlined by Törnquist’s framework. This capacity is comprised of four elements: actors’ activities in various parts of the political terrain and their capacity to link these; their strategy of politicizing issues and interests; their strategy of mobilizing people; and the way they relate to institutions and rights. Afterwards, I assess other conditioning factors on actors’ capacity; actors’ political opportunity structure, their habitus, and their sources of power, i.e. various forms of capital, as well as their ability to transform these into symbolic capital. In order to make clear what the assessment of these factors, as well as actors’ capacity, and the quality and scope of institutions and rights is about to do, I outline the precise questions being asked below. The aspects by Crook and Manor are included in the questions on the accountability of local government and state administration:

-Do local government, as well as a local state administration exist, which are open, accountable, and representative? Do the judiciary, rule of law, and citizenship exist? What is their quality and scope in terms of ensuring equal treatment before the law 47 and reaching all people in the municipalities? Are all civil and political rights respected? Does a democratically oriented civil society exist? Were municipal elections in 1998 free and fair? Who are the political actors in each case? Do they have the capacity to link activities in different parts of the political terrain, politicize issues and interests and mobilize people? How do they relate to existing institutions and rights? Do their opportunity structure, their habitus or the forms of capital they possess contribute to explaining their capacity?

7.1 State of affairs of institutions and rights 7.1.1 State administration In this section, I assess the existence, quality and scope of the state administration at the local level, including village/ward chiefs. The municipal administration is part of local government and is examined in the following section. The state has neither transferred all responsibilities to the local government nor deconcentrated its administration yet. However, it has devolved some tasks, and thus reduced its position locally to one of tutelage through controlling a posteriori that decisions taken abide by the law. Additionally, the state representative (sous-préfet) participates on municipal council meetings when asked to by the council. In all three municipalities, the state performs these tasks. Still, because of the weak state presence in Kontela and Tambaga after the reform, the changes have not been great. The state is more absent in Tambaga than in Kontela, because it does not have an office in Tambaga. People only feel the state’s presence when guards intervene to accompany tax collectors (the accountant and village chiefs) forcing people to pay (municipal employee 28.10.03 [interview]). Furthermore, the state has lost legitimacy in both cases, judging by the level of tax avoidance. Neither in Kontela nor in Tambaga has the state managed to provide services for the population (state employee 14.10.03 [interview]; République du Mali:28; councilor 2004 [e-mail]). Moreover, the state has misused tax revenue in both cases (anonymous 17.10.03 [interview]; Tag 1993:39). The quality of the state administration in both cases is thus bad. The only people in both cases having contact with the state are the local government and the village chiefs. Because of the weak representation of the state, its 48 scope is rather narrow. In Mopti, though, the state has a substantial presence, which causes administrative employees to be skeptical towards decentralization, because they loose some of their privileges and influence because of the reform (state employee 11.11.03 [interview]). The administration is engaged in clientelistic practices in sales of property, whereby citizens have to pay state employees to process the necessary paperwork (Bouju 2000:156-159). Such practices greatly reduce the quality of state services, and the state’s legitimacy in the population. The state administration’s scope is however wide, because state services are available to the population of the municipality and they may easily get in contact with the state. The record of tax collection, which is even worse in Mopti than in the other two municipalities, reflects the lack of legitimacy. The record is worse, because it is more difficult to control the payment of taxes in cities than in rural areas (state employee 11.11.03 [interview]). The difference in state presence between rural and urban municipalities applies to other African countries as well (Bierschenk and Sardan 1997, Sawadogo 2001). Still, if one includes the village chiefs in the state administration, the quality and scope of the state are greater in all three cases. In Mopti, the village chief is the chief of the ward chiefs and he enjoys a symbolic and informal position resembling that of the village chiefs in the other two cases. The ward chiefs, however, are weaker, because they do not have the same status of custom bearers, and therefore people do not respect them as much (state employee 11.11.03 [interview]; Massing 1996a). In all three cases, village/ward chiefs represent the population at meetings with the state and municipality. In Kontela and Tambaga, they also mediate in conflicts. Nevertheless, they do not perform their most important responsibility: collecting taxes. This is due to people relating to them as customary authorities rather than state representatives. Hence, their position in the population decreases the quality of state administration, because the lack of tax revenue impinges on state transfers to the three municipalities.

7.1.2 Local Government- Municipal Council and Administration In this section, I assess the quality and scope of local government, i.e. the municipal council and administration in terms of their open and representative character, which, 49 for the purposes of this assessment, includes the village/ward chiefs, because of their consultative role and influence over local government. In Kontela, the municipal council hardly functions (state employee 11.10.03 [interview]; anonymous 17.10.03 [interview]). Its scope is thus narrow and its quality low, owing to several factors. Firstly, most of the municipal councilors do not know what their tasks are, because they have received no information about them (state employee 14.10.03 [interview]). This makes it difficult to understand the work of the council. Secondly, few councilors can read, and in any case, the few documents that exist are in French, which few councilors understand. Thirdly, the dominance of the deceased mayor hindered all previous proposals by other councilors (Amundsen2 04.02.04 [interview]; anonymous 17.10.03 [interview]).Fourthly, some people in local government abuse tax revenue and revenue from conflict mediation in Kontela (anonymous 17.10.03 [interview]; anonymous 2004 [e-mail]). Fifthly, it is both a cause and a consequence of the council’s small output that it is of little importance to the population in Kontela. Few people even know that it exists, according to several informants. The will of the nobles, a majority on the council, is followed. People expect the nobles to decide, because they are the existing political elite in the villages (anonymous 17.10.03 [interview]). The representation of only nobles and the two ex-slaves on the council excludes all other groups, including women and youths, from decision-making (CCC Bafoulabé 2001). Because of the strong role of the nobles in decision-making, the government is not representative and open, which is another cause for its narrow scope and low quality. Decisions on the council are usually taken by consensus, but if discussions take too long, they vote (councilor, 09.10.03, [interview]; state employee 11.10.03 [interview]). The village chiefs influence decisions through consultations and other informal talks before meetings (state employee 11.10.03 [interview]). None of the commissions established – education, culture, property, and civil matter registration- functions, due to lack of knowledge and organization. The general

2 The informant is a former missionary in Kontela for the Norwegian organization Normisjon from 1997-2003. 50 secretary and the accountant are the only two municipal employees working on a daily basis, and this is one of the reasons for why the local government has not been able to do much for the population. As to the government of Tambaga, it has higher quality than Kontela’s, because it has achieved more in terms of services, due to having some skilled councilors, among them, the mayor and a competent administration. These actors have been able to retrieve funds from ANICT, and to make a detailed development plan, which is required in order to get funding (Soumonou 23.11.03 [interview])3. The scope of the government in Tambaga is wider than Kontela’s, because the government fulfills people’s wishes to some extent, even though there are only nobles on the council. It is thus more representative than Kontela’s council, but less open, since only nobles are on Tambaga’s council. However, a misunderstanding on the council in the first year reduced the government’s quality. This misunderstanding was that some councilors refused to participate in council meetings, because they thought that the council was only for people belonging to the political party of the mayor (councilor 28.10.03 [interview]). Additionally, conflicts on the council have made collaboration difficult. The two rural cases and Mopti case are poles apart, because of the long experience of local government in Mopti. Further, Mopti’s councilors have high education and a good command of French (NGO worker 10.11.03 [interview]). Also, the administration in Mopti has about a hundred employees in technical services, health and so forth, and they serve the population through decentralized centers in five of the wards, and its quality is thus high. Contact between the population and the administration is frequent owing to the services, which makes the government’s scope wide, reaching the population. The sale and distribution of property, however, reduces government’s quality, because the council operates with two prices in such sale; the unofficial (higher) price, which is the price actually paid, and the formal, which is lower. Municipal councilors pocket the difference (Massing 1996a:23-24). Also, every time a group of lots is sold,

3 Soumounou works for ANICT at the head quarter in Bamako. 51 councilors get five lots each (ibid.:23), and the council give priority to the village chief and the founding families in the sale(Massing 1996a:20; state employee 11.11.03 [interview]). Due to these practices and the corruption by the state, described in the latter section, the population has lost confidence in the municipal council (state employee 12.11.03 [interview], state employee 11.11.03 [interview]; Bouju 2000:157; Massing 1996a). Hence, the government is not representative of people’s interests in their work. Nor does the government give an account of how it has used tax money, which reduces the quality of government (state employee 11.11.03 [interview]. Furthermore, through informal discussions late at night between councilors and other actors, excluding the female councilors, councilors make concessions and agree upon positions before the council officially meets for decision-making (state employee 11.11.03 [interview]). Therefore, government is not open. As to the roles of village/ward chiefs, they differ between cases. In Kontela, they are sometimes consulted, but consultation is not as widespread as in Tambaga where “the mayor always consults them before the council discusses issues and decides on them” (councilor 2004 [e-mail]).The influence enjoyed by the village chiefs is considerate, but it varies, because not all village chiefs participate in every meeting with the council (councilor 28.10.03 [interview]). In Mopti, the village chief and the ward chiefs reduce the quality and scope of the council and administration, because these chiefs make people refuse to participate in those municipal projects that the chiefs oppose (state employee 11.11.03; NGO worker 13.11.03 [interview]). Consequently, they hinder the council from being accountable and representative (Bouju 2000). Despite this sanction power, the village chief and ward chiefs complained that they had lost influence since 1992, because the municipal council had taken decisions against their will and had not consulted them in the decentralization process (Massing 1996a:20). Still, they influence politics, through neo-patrimonial relations with the council (Bouju 2000). (Bouju calls these relations clientelistic, but I refer to them as neo- patrimonial when they involve politics). The ward chiefs, however have less influence than the village chief has, because the population regards them as less 52 important, and many do not even know who the ward chief of their ward is (state employee 11.11.03 [interview]). However, according to Bouju (2000:154), they are still part of clientelistic relations and important in elections, because they mobilize voters. Altogether, ward chiefs thus wield political influence, but to a lesser extent than the village chiefs. Part of their influence relates to that they cooperate with the municipal administration on administrative matters. Although Mopti’s village chief and ward chiefs influence politics, they would never engage themselves in political parties (Bouju 2000), as some of the chiefs in Tambaga do (councilor 28.10.03 [interview]).

7.1.3 Accountability In this section, I will assess accountability based on the framework of Crook and Manor, by studying two aspects: participation and institutional performance. Several variables measure these two aspects.

Participation

The first variable is direct participation, which includes participation in formal and informal meetings related to the municipal council’s work. In all three cases, participation in such meetings varies greatly between different parts of the population, due to age, gender and the caste system. For instance, village chiefs in Tambaga frequently take part in meetings with the municipal council. Meanwhile, the council sometimes holds meetings, which are open to the public, but to males only. In these meetings, they give an account of municipal spending and ask the population about its needs (councilor 28.10.03 [interview]). Additionally, Tambaga’s village chiefs hold informal meetings with heads of families to inform them of the council’s decision. The village chiefs then expect the heads of families to inform their families. Although, all the meetings mentioned exclude women and young people, from time to time some village chiefs call upon women’s and youth associations in matters concerning them (village chief 30.10.03[interview]). The mayor also consults the women’s associations sometimes (woman 29.10.03 [interview]).Turning to Kontela, the only meetings there are between the mayor and village chiefs, but these are 53 irregular meetings. The mediation of disputes at the city hall thus constitutes the most frequent contact between the municipal council and the administration in Kontela, as well as in Tambaga. In Mopti, people are frequently in touch with the municipal administration, but not with the ward chiefs (state employee 11.11.03 [interview]). There is, however, regular contact between the municipal council and the ward chiefs, especially when the municipal council makes budget plans (ibid.). The council recognizes that the ward chiefs know the needs of the population. Therefore, there are no meetings between the wider population and the municipal council. Rather, the council informs the population about such matters through radio, posters, and newspapers. The second variable of participation, the relationship between the electorate and the council, measures contact between individual voters and councilors. According to several informants, the council only maintains contact with the electorate in Kontela during campaigns in municipal elections, while in Tambaga the council holds information meetings on a regular basis. In Mopti, the municipal council maintains contact through the press, and also through the village chief and ward chiefs. Participation by marginalized groups is the third variable of participation. It assesses participation by women, youth, and lower castes. Participation by such groups varies between the municipalities as well. In Kontela and Tambaga, the pattern of participation at the village level is identical to the pattern of participation in the municipal council as a consequence of the nobles participating in both councils. It seems like the population expects the nobles to transfer their existing participation and power position from the village council to the municipal council. Still, the similarity of patterns of participation at the village and municipal level does not mean that the situation pleases the population. The dominance of the nobles simply shows that it is difficult to change the perception of the nobles as the governing elite. Consequently, decentralization has not improved representation by other social groups. However, together with the multiparty system, decentralization has made a difference in participation in political parties. In parties, all castes take part, but not other ethnic groups in Kontela and Tambaga, according to several 54 informants. In Tambaga, even women participate, on separate party committees. The activity level of the parties varies; in Kontela, they are only active during elections, while in Tambaga they are active at other times as well (councilor 28.10.03 [interview]). In Mopti, three features make political participation different from such participation in Kontela and Tambaga. Firstly, clientelistic relations between founding families, traders and poor people condition social relations and political participation, and not the caste system. These relations constitute a safety net for poor people (Bouju 2000), and they are important for poor people’s survival as well as in politics, replacing the support, which the state should provide to the population. Such relations are hard to break. The significance of these relations for the poor elucidates why they do not mobilize in politics on their own. In political participation, clientelistic relations turn into neo-patrimonial ones, because if a candidate is not a member of the founding families, or the village chief’s family, he has to engage in or maintain neo-patrimonial relations with these actors in order to gain a seat on the municipal council (Bouju 2000). The tactic by the founding families is to get their clients and family members elected to the municipal council. The aim of this tactic is to ensure control over the property market in the city, over which the municipal council governs (NGO worker [interview]). Consequently, it is important for these actors to maintain relationships with politicians, because the politicians have access to and power over distribution of resources (NGO worker [interview]; Bouju 2000). To participate in politics in Mopti, it is thus not sufficient to use the caste system, as in Tambaga or Kontela. Rather, neo-patrimonial relations in Mopti replace the caste system in determining political participation. The many ethnic groups in the municipality, that each has a noble class, explain the inadequacy of using one’s status in the caste system for political participation in Mopti. Secondly, the only way to participate without using these relations is to have a high position in a political party (Massing 1996a:25). The parties nominate the most influential members as council candidates (Massing 1996a) Nevertheless, it is difficult to get the population to be active in political parties, and therefore, parties do not represent the will of the people (state employee 11.11.03 [interview]; Bouju 2000). Thirdly, age and gender do not 55 condition participation to the same extent in Mopti as in the other two cases. Concerning gender, three women are councilors. It has been difficult to assess women’s participation in general, but according to informants, women mobilize more than men do in Mopti, but they do not partake in informal political meetings (see sect. 7.1.2) (state employee 11.11.03 [interview]). Regarding age, many of the councilors are quite young (ibid.). Youth therefore participate more here than in Kontela and Tambaga. The fourth variable measuring participation is whether the municipal councilors are known to the population. I have incorporated the aspect of information from the council to the electorate in this variable, building on Blair’s (2000) assessment of accountability. In Kontela, few people know that the municipal council or the councilors exist (anonymous 17.10.03 [interview]). In Tambaga, some people are familiar with the councilors. Most people know about the municipal council and the councilors in Mopti, since the council has been there for many years (NGO worker 10.10.03 [interview]). As to information, Kontela’s municipal council does not inform the population (state employee 14.10.03 [interview]). The councilors have not understood the significance of doing so. Tambaga’s council sometimes holds public meetings to inform the population about decisions taken. Mopti’s population gets more information of the council’s work than the populations in Kontela and Tambaga do, by radio, newspapers, and posters, and through the ward chiefs. Compared to the other two municipalities, Mopti’s council informs people well. However, they only provide partial information, because they do not reveal the bargaining that is part of decision-making.

Institutional performance

Regarding institutional performance, the two dimensions of output efficiency and process assess it. The first variable of output efficiency is the responsiveness of local government towards the electorate. This also relates to the second variable of output efficiency: to compare local governments’ achievements to the previous government in Mopti. For Kontela and Tambaga it implies comparing the state administration to the government. The state still takes responsibility for devolved services in most 56 municipalities by financing them. These services are in the domains of education, health, water, and registration of civil matters. Consequently, the state contributes to enhance the responsiveness of municipal councils in Mali. As to registration of civil matters, this does not work in Kontela and Tambaga, because people do not register themselves (municipal employee 16.10.03 [interview]). In Mopti, however, they do register. The reason is probably that people have got used to it, since the municipality has handled civil matters for years. Regarding education and health, the situation also differs between cases. Kontela’s councilors are concerned about education and health, but not able to build any schools or provide health services, because they lack knowledge of how to get state funding (state employee 14.10.03 [interview]; municipal employee 23.10.03 [interview]). The little tax revenue also contributes to the poor responsiveness in services. The few improvements in these sectors result from the emigrant associations, cooperating directly with the population or from the population themselves (municipal employee 16.10.03 [interview]). Despite the few improvements by Kontela’s council, it has achieved more than the previous state, which did not provide almost any services, according to several informants. Kontela’s difficult situation regarding all public services is thus partly due to the previous state administration’s lack of work. Tambaga’s municipal council is more responsive than Kontela’s, because it has improved the situation in public services for the population. Compared to the previous state administration it has also achieved more, because the state provided few services in Tambaga, according to several informants. Through financial backing from the state agency, ANICT, Tambaga’s council has financed schools, a city hall, and recruited two midwifes. Skilled councilors that know how to apply and receive funding explain the difference in responsiveness between the two cases. Mopti’s municipal council is more responsive than the other two, because it has improved services in many domains the last years, as in education (municipal employee 12.11.03 [interview]). However, the services the previous local government and the state established before decentralization is also an important explanation for Mopti’s better responsiveness, and thus higher accountability. Due to these services, the council has expertise that it 57 profits from in improving existing services. Still, despite these achievements and provision of numerous other ones, Mopti faces persisting challenges in many areas. Further, none of the three municipal councils has fulfilled people’s needs, and been sufficiently responsive. Lack of revenue constitutes the most significant reason for insufficiency. This deficiency relates to the second variable of output efficiency: people’s opinion about the council and administration. In all cases, they do not trust these institutions. Hence, they do not pay taxes. They believe the state or mayor will “eat” tax money (councilor 28.10.03 [interview]). The population’s distrust is a consequence of their experience with corruption by the state since independence (Lange 1999). They transfer this suspicious attitude towards the state to municipal councils. Tambaga and Kontela’s lack of tax revenue also results from people still believing that the state has abolished all taxes, although the state has run information campaigns to clear out this misunderstanding (see ch 4). These campaigns have not reached the population of the two cases, which shows the state’s absence in the two cases. In Mopti, people do not pay taxes, because the local government does not give an account of on what they spend tax money. This non-transparency in public spending has led to a“confidence crisis” between the population and the municipal council (state employee 12.11.03 [interview]). From output efficiency, it is time to turn to process, the third dimension of institutional performance. Process includes practices that may affect accountability, such as corruption and rent-seeking. Rent-seeking and corruption is present in Mopti’s neo-patrimonial relations, and in mediation in disputes in Kontela, previously described in this chapter. These practices weaken the two local governments’ accountability, because they reduce the quality of the two governments’ institutional performance.

Opposition parties

Opposition parties may be “a powerful engine for enforcing accountability” (Blair 2000:28). However, the political parties in Kontela and Tambaga do not fill this role, because none of their members has an ideological base. People have engaged 58 themselves in parties to gain money or contacts, according to several informants, and many do not even know what the parties stand for. Due to this lacking awareness, some councilors changed parties, and it was thus difficult to build up a stable opposition on the councils. Further, most decisions are taken by consensus, as described in section 7.1.2, and parties do therefore not matter much. At least in Kontela, personal power matters more than party loyalty (Amundsen 04.02.04[interview]). In Mopti, personal interests mattered more than political parties (Bouju 2000:152). Still, some politicians were also active on ideological grounds, but these did not form an opposition force enforcing accountability. Altogether, the situation in the cases is thus similar to Blair’s conclusion (2000:31) that political parties are no accountability mechanism in Mali. Summarizing the entire discussion of accountability, Kontela’s local government is not accountable, because its institutional performance and participation by the population remain low. Tambaga’s local government is somewhat accountable, owing to higher institutional performance and more participation than in Kontela. Mopti’s local government has higher accountability, because its institutional performance and participation are superior to the other two cases.

7.1.4 Free and fair elections? This section assesses the quality and scope of municipal elections in 1998 and 1999, in order to determine whether they were free and fair in the municipalities. Overall, municipal elections held in Mali in 1998 and 1999 were free and fair (UNDP 2003:13). Yet, some elements had an impact on people’s voting in the three cases. Even if these elements did not affect the freeness and fairness of the elections, they may still have influenced results, and therefore merit study. These elements are described below. As to Tambaga, one element that may have influenced elections was that many people refused to register to vote, because they thought elections were only for politicians. In addition, in Tambaga, people only considered nobles eligible. Hence, local perceptions of eligibility determined voting and thus posed restraints. Several informants in Tambaga also stated that many husbands dictated their wives’ choice. Additionally, some village chiefs in Tambaga ordered the population to vote 59 for one particular party. Also, people in Tambaga voted at the village chief’s residence. The location of voting offices in such residences may have influenced voting, because some village chiefs were political party members, and their residence was therefore not a neutral site. Turning to Kontela, the voting offices there were also mostly the village chiefs’ residence. In Kontela as well people thought nobles were the only ones eligible. Another factor that may have affected elections in Kontela was long distance to voting offices. Regarding Mopti, neo-patrimonial relationships between the electorate and the village chief, the ward chiefs, and founding families affected elections. These actors constituted one group in the relationships, and functioned as ”grands electeurs” (voters influencing others) (ibid.:153; Ouane 21.11.03 [interview]). In elections, “the electorate is expected to satisfy the village chief that it has received services from” (Bouju 2000:15).

Hence, people in general “rarely vote on political conviction”(ibid.:152), but to preserve their interests (ibid.), i.e. clientelistic or neo-patrimonial relations. Despite the village chief, ward chiefs and founding families wielding influence, they would never present themselves as candidates. Their strategy was to engage in neo- patrimonial relations with all political parties, and never spelled out their political preferences (ibid.). Due to this strategy, they influenced several political parties. There was thus a vote-catching relationship between different groups. According to Beridogo (2002), this relationship is common in Malian politics. Altogether, the obstacles to voting discussed here influenced electoral participation and possibly, election results.

7.1.5 Civil and political rights, citizenship and the judiciary This section assesses the existence, quality and scope of citizenship and the judiciary, as well as the five civil and political rights listed in chapter five. In Mali, there are few abuses of civil and political rights (Sandbrook 2000:38). In the cases, however, the right of equality before the law is violated to some extent. The local government act has institutionalized use of customary law in Mali. The state does therefore not guarantee people’s equal treatment before the law. Consequently, citizenship does not fully exist. The village chief/ward chief and the municipal council, primarily the 60 mayor, constitute the judiciary institution at the municipal level. According to several informants in Kontela and Tambaga, village chiefs and the municipal council apply customary law here. In most instances in Kontela and Tambaga, the village chief and his council mediate first, and transfer it to the municipal council if they cannot handle the dispute. In Kontela and Tambaga, the arbitrators on the council make the parties pay for mediation and profit on it (state employee 14.10.03 [interview]; anonymous 17.10.03 [interview]). Because of the extensive use of customary law in Kontela and Tambaga, a judiciary system ensuring equality before the law does not exist. This also implies that citizenship has low quality and narrow scope, only present for those that are able to go to the district courts to try their cases. Regarding the other civil and political rights, they exist and have good quality and wide scope, with the exception of the freedom of assembly and organization in Tambaga. The village chiefs violate this right, because they require political parties to seek their permission to organize political meetings (municipal employee 28.10.03 [interview]). Turning to Mopti, the village chief uses customary law, while the council/mayor applies both this law and written, unified national law (ref). Mediation is conducted predominantly in matters concerning natural resources, and leaders for different associations of natural resources act as mediators in addition to the village chief (NGO worker 13.11.03 [interview]). In Mopti, the civil right of equal treatment is present to some extent, since the mayor/council applies written national laws. However, although customary law may treat people equally and fairly, such treatment is not guaranteed, because customary law comprises fluid principles that change over time. Regarding the other civil and political rights assessed, they are fulfilled to such a level that the quality of civil and political rights is fair and their scope wide.

7.1.6 Civil society In this section, I assess whether a democratically oriented civil society exists in the cases. First, I examine whether civil society exists in the cases, and if it exists, I will describe its make up. Secondly, if it exists, I will measure civil society’s quality and scope by these two elements: people’s participation in civil society and its influence on local government. Altogether, existence, quality and scope assesses whether the 61 cases’ civil society is democratically oriented. In Kontela, civil society exists, because several women’s associations (women only), men’s associations (men only), youths’ associations (both sexes), and hunters’ associations (men only), as well as two emigrants’ associations are present. The emigrants’ associations consist of emigrants from the municipality living in France, but they also have local members administrating their work in the municipality. As in many other West African countries, most associations engage in agriculture. The activities of the women’s, men’s and youth’s associations are collective agricultural and handicraft work, whereby they accumulate savings. The associations function as banks to use the accumulated savings as a safety net for members. Further, the emigrants’ associations are Kontela’s most important organizations, because they are the only ones able to initiate and/or finance development in the municipality. Kontela’s civil society has wide scope, because most of Kontela’s population participates in one or more of the associations mentioned. However, it has a low quality, because only the two emigrants’ associations have an impact on local government. Their impact is due to that they provide necessary financing for development projects in the municipality. Because of civil society’s lacking influence on Kontela’s government, its civil society is not democratically oriented. In Tambaga, a civil society also exists. It is composed of the same associations as described above, with the exception of emigrants’ association. Still, here the women’s associations have some influence on the council and administration. That the mayor sometimes consults them indicates their influence (woman 29.10.03 [interview]). However, a democratically oriented civil society is not present here, either, because in general, few associations influence municipal council. Mopti’s civil society constitutes a democratically oriented civil society to some extent, because some of its associations have an impact on the council, described later on. Mopti has the same associations as Kontela and Tambaga (except the emigrants’ associations).Unlike the situation in the other two cases, several of Mopti’s associations have officially registered as associations (NGO worker [interview]). This allows them to seek external financial support, and they have more activities than in 62 the other two cases. These registered associations make civil society wider in scope than in the other two cases, because they have more activities because they are externally subsidized. Also, civil society’s scope is wider than in the two cases, because several other associations, organizations, unions, and other groups exist. One of these are the economic interest groups (GIEs) which perform public services, such as waste collection and levying and collecting taxes on the market (state employee 11.11.03 [interview]). In addition, primordial associations partly based on the caste system and partly on ethnicity, exist. These associations manage natural resources and resemble trade unions, because they organize and assemble socio-professional groups/castes, such as anglers, artisans, and cattle breeders. Some of these groups consist of several ethnic groups, while others comprise only one (Massing 1996a:16-17). Within each natural resource, there are several associations/groups with one leader each (ibid.). Together, all these associations engage large parts of Mopti’s population. As to whether the associations have an impact on local government, the natural resource associations do, in the domain of natural resources, because they assist local government in natural resource management (Massing, 1996a:21, state employee 12.11.03 [interview]). According to several informants, their leaders have a particularly dominant position towards local government, since they often mediate in local resource conflicts, which are numerous in Mopti. Other groups influencing local government are the economic interest groups that perform public services. By their work, these groups may contribute to enhance the responsiveness of the council and administration towards the population. Some of the women’s associations also have an impact on the council, but their influence depends on their capacity to understand the political game, having a strong leader and good internal organization (NGO worker [interview]). All in all, Mopti’s civil society is to some extent democratically oriented, since some groups influence local government. The reason for civil society only begin somewhat democratically oriented is that the most powerful associations, the natural resources’ associations only engage some population groups. Nonetheless, Mopti’s civil society has higher quality and wider scope than Kontela’s and Tambaga’s. 63

7.2 Actors’ capacity to use institutions and rights This section will assess political actors’ capacity to use institutions and rights in the three cases. The political actors considered here are the ones that are active in politics in the three cases: municipal councilors, village chiefs (and founding families in Mopti), state representatives (state administration in Mopti), members of political parties, and civil society. Mopti’s ward chiefs are not included, because they do not wield as important an influence as the village chief and founding families. Capacity is assessed by looking at four elements: 1) Actors’ activity in various parts of the political terrain and their capacity to link these; 2) Actors’ politicization of issues and interests; 3) Actors’ mobilization of people; and 4) Actors’ relation to institutions and rights (Törnquist, 2002,2003). The latter depends on element two and three. Most of the information employed in this section has been presented in previous sections of this chapter. Therefore, several references to interviews are left out here. I measure each of the actor’s capacity by these four elements in turn, starting with the cases of Tambaga and Kontela. These cases are studied together, because they have many similar traits. I start off discussing the capacity of Kontela and Tambaga’s councilors regarding the element to link activities in different parts of the political terrain. Councilors are able to do so, because they are simultaneously present in the political parties and the government sections. Concerning the second and third element assessing capacity, actors’ politicization of issues and mobilization of people, Kontela and Tambaga’s councilors politicize single issues, such as personal interests. The purpose of politicizing interests/issues is to gain resources; i.e. property or money. By politicizing these single issues, they politicize communal interests and loyalties at the same time, since the money/resources benefit their families and networks. By this politicization, they are thus able to mobilize people. This mobilization crystallizes in elections, where they mobilize people with gift giving and by using the loyalty bonds of the caste system. Altogether, their mobilization strategy is thus incorporation of people (see sect 5.1). According to Törnquist’s framework, their strategies of politicization and mobilization together constitute a joint strategy, which lies in 64 between populism and clientelism. Nevertheless, some of the councilors also have collective interests, such as contributing to development of the municipality. As to the councilors’ politicization towards the village/ward chiefs, they politicize by communal loyalties, because they appeal to the village/ward chiefs’ important position in the villages/wards that has endured for centuries. Further, they mobilize these chiefs by incorporating them in decision-making and administration of the municipality. However, Tambaga’s council does this more than Kontela’s. Their joint strategy towards the village/ward chiefs is thus in between networks and organized integration. It tends more towards networks, because of the long-term social role of the chiefs on which the council’s mobilization of them builds. Regarding the fourth element, the way actors relate to institutions and rights, most of Kontela’s and Tambaga’s councilors use them. They use them to further instrumental aims, such as to gain resources and contacts with higher level politicians. Still, some councilors are also abusers, such as the ones changing parties in the mayoral election in Kontela. Still, they do not have much capacity to abuse institutions and rights, because there are few resources in the two municipalities. Moreover, there are some pro-democrats in both cases. Tambaga’s few pro-democrats have the capacity to promote institutions and rights, while Kontela’s do not. The many improvements in public services in Tambaga compared to Kontela show the difference in pro-democrats’ capacity the two places. Kontela’s and Tambaga’s state representatives are only present in the local state part of the terrain. Consequently, they have no capacity to link activities. They try to politicize collective interests, by appealing to people to pay taxes, but are not successful (state employee 11.10.03 [interview]; state employee 31.10.03). They only mobilize the councilors when the latter ask them to give advice, or when they control that the councils’ decisions abide by the law. Additionally, they integrate village chiefs, since they are part of state administration. The state representatives’ overall aim is to preserve national unity, and the mobilization strategy is thus integration. According to the framework, the state’s strategy of politicization and mobilization combined thus lies in between networks and integration, but the state is much closer 65 to integration than networks, because the municipalities are part of the Malian state system. Concerning the state representatives’ relation to institutions and rights, they are pro-democrats on one hand, because they assist the municipalities and thus promote institutions and rights. On the other hand, they have also used and abused institutions and rights, by resisting the reform, in order to promote their single interests of preserving their pre-decentralization dominant positions locally. This resistance has further made them anti-democrats, because they have not provided information about the reform needed by the local governments and the populations. They have thus neglected their responsibility, and undermined institutions and rights. Concerning civil society, it does not have the capacity to link activities in various parts. Regarding politicization, all associations politicize collective interests. Further, all associations mobilize people, because most of Tambaga’s and Kontela’s populations take part in association activities. Their mobilization strategy is therefore of integration. For instance, both emigrants’ associations in Kontela mobilize by integrating parts of the local population in development projects (municipal employee 16.10.03 [interview]. The joint strategy of politicization and mobilization thus lies in between networks and organized integration. As to civil society’s relation to institutions and rights most of the associations are preoccupied with economic activities. Hence, they do not use, abuse or promote institutions and rights, with the exception of the emigrants’ associations in Kontela, which promote democracy by providing public services, and thus help the municipality achieve accountability (République du Mali). Still, until recently they bypassed local government to get their projects done, because they did not trust it (anonymous 17.10.03 [interview]). Political parties, however, link activities at national and local levels. The mayor’s election in Kontela demonstrated this capacity (see sect 6.1). Concerning their capacity to politicize, most party members have single interests in mind: getting access to resources and contacts. Their constant changing of political parties reflects these interests. However, communal loyalties are also present, because politicians appeal to caste relations and family bonds. Besides, politicians mobilize the population by giving gifts during election campaigns, and trying to make them 66 become members of their parties. Their mobilization strategy is thus one of incorporation, while their joint politicization and mobilization strategy lies on the continuum between clientelism and populism, tending towards clientelism, because gift giving is used more than populist arguments. As to their relation to institutions and rights, members of Kontela’s and Tambaga’s political parties use them, because they engage in parties to gain resources, contacts, and influence. There is nothing wrong with having such instrumental aims, as long as party members do not end up abusing institutions and rights to achieve these aims. However, abuse actually takes place, in elections by giving gifts and by appealing to caste bonds. Regarding village chiefs, they link activities in the self-government, the government, and the political societies sections of the political terrain, and thus have great capacity. Moreover, they politicize on single interests and collective interests, and communal loyalties, because they politicize both on their customary power status and on representing the village. Their strategy is to maintain their power. The chiefs incorporate people when they mobilize them, such as by using reciprocal relations between castes (see sect 4.4). Consequently, the village chiefs’ joint politicization and mobilization strategy lies in between clientelism and populism. They tend towards clientelism, since they have reciprocal relations with the population. Moreover, they are pro-democrats on one hand, because they contribute to the municipal councils’ accountability. This is because they represent people’s needs in meetings with the councils. Still, on the other hand, they are anti-democrats. The chiefs undermine institutions and rights, because they do not fulfill their responsibilities in tax collection, and in Tambaga they restrain the councils from implementing decisions if they do not agree with them (municipal employee 28.10.03 [interview]). Turning to Mopti, its state administration has no capacity to link activities in different parts of the terrain, since it is only in the government section. Next, Mopti’s state is able to politicize on single issues, by engaging in clientelistic relations in property management (Bouju 2000). Hence, it is thus able to mobilize people by state employees engaging in clientelistic short-term relations with people, described in section 7.1.2. The mobilization strategy is thus incorporation. The state 67 also tries to politicize collective interests, appealing for tax revenue (state employee 12.11.03 [interview]). The joint strategy of Mopti’s state administration lies in between clientelism and populism, but tends towards clientelism, due to the clientelistic relations described above. Further, the state employees are anti- democrats, exemplified by the resistance towards decentralization among them. However, there are also pro-democrats in the state administration because they try to help the municipality achieve development. Concerning Mopti’s municipal councilors, they combine activities in the government and political societies parts through neo-patrimonial relations with the electorate, the village chief, and the founding families. As in the other two cases, they politicize on single interests, mostly property. They mobilize the population and the village chief through neo-patrimonial bonds, thus a strategy of incorporation. This happens in for instance sales of property. The municipal council gives priority to the village chiefs and founding families in the sales of property. These actors may again distribute property or profit earned on selling these properties to their clients. In return, councilors get support in elections and the population gets help to survive. This practice is not present in Kontela and Tambaga, because there are few resources there. Even if the mobilization strategy of Mopti’s councilors is by neo- patrimonial relations and not clientelistic ones, their joint strategy lies on the continuum between clientelism or populism. This is owing to that Törnquist argues that his definition of clientelism also includes neo-patrimonial strategies (ibid.:39). However, the joint strategy of Mopti’s councilors is more towards clientelism, because the relations established with the electorate and village chiefs and founding families endure for a longer time. Due to this strategy, Mopti’s councilors abuse institutions and rights. However, there are also some pro-democrats, because the council has taken decisions that benefit the entire population and not only the village chief and the founding families. The political parties that have representatives on the municipal council have the capacity to link activities in government and political societies. Further, most political party members politicize on single interests. They use neo-patrimonial relations to mobilize the population and the village chiefs and 68 founding families, which results in an incorporation strategy. Together with politicization on single interests, this leads to a joint strategy lying on the continuum between populism and clientelism. The strategy is closer to clientelism than populism, because political parties use neo-patrimonial relations that persist over time. By these relations, they use institutions and rights to further instrumental aims. Mopti’s political party members are as in the other cases mostly consumers of democracy. However, whenever they distribute property or other resources to their neo-patrimonial relations, their use turns into abuse, and they become anti-democrats. Moreover, unlike the other two cases, some of Mopti councilors are pro-democrats, because they have engaged themselves on ideological or collective interests, and have been able to promote institutions and rights through fulfilling people’s needs to some extent. Altogether, however, most of the councilors are users and abusers of institutions and rights. As to Mopti’s village chief, he links activities in the self- government unit, the political societies, and the government/administration section at the same time. This presence in the latter is due to his position as an administrative state representative. The founding families, however, are not in the self-government section, because they do not have such a role as the chief does. Still, both the village chief and the founding families are in the political societies, because they have neo- patrimonial relations with political parties. Further, concerning their politicization, they have single and collective interests, and communal loyalties as in the other cases, but these are restricted to their enlarged family and the clientelistic and neo- patrimonial network in which they are. The network serves their single interests. Hence, they mobilize by incorporation. Their joint strategy is in between clientelism and populism, but tends towards clientelism, because their neo-patrimonial relations are stable over time. In the other two cases, these relations only exist during election campaigns. Further, Mopti’s village chief is both a pro-democrat and anti-democrat. He is an anti-democrat, because he utilizes his customary power to wield influence, and hinders the council’s implementation of decisions, and hence its accountability. The founding families are also anti-democrats, because they have power to sanction 69 the municipal council’s decisions. However, only the village chief is a pro-democrat, due to his role in representing the population’s views on different matters to the local government. As to civil society’s capacity, some associations here are in two parts of the political terrain; the civil societies part and the government part. These are the associations, which perform public services, and the natural resource associations that mediate in conflicts for the local government (see sect 7.1.5). They politicize on collective interests; the municipality’s development and well-being. Large segments of the population has organized itself in such associations (councilor 11.11.03 [interview]), and thus the mobilization strategy of these associations is integration. Their joint strategy lies in between networks and organized integration. The specific strategy varies from association to association. The natural resources associations, for instance, tend towards stable organized integration, since these are associations where membership is mostly based on belonging to a caste or ethnic group, which do not change over time. Further, Mopti’s civil society consists of some pro-democrats. These pro-democrats are the ones that put pressure on the municipal council in various matters. Hence, they make the council work to achieve development. For instance, the associations performing public services are pro- democrats since they help the council obtain accountability. It is, however, likely that there are abusers in Mopti’s civil society as well, trying to profit from these associations’ resources. No documentation is present on this issue, because it is sensitive, but it is highly probable that some abuse exists, due to two reasons. The first is that resources are the motivation of other political actors, namely councilors, and is it likely that some actors in civil society also have this motivation. The second reason is that there is abuse in other West African countries (Blundo 1995). As to the population in general, it does not have any separate capacity to link activities in any of the cases, nor does it politicize the other political actors or mobilizes outside being part of the different groups of political actors discussed here. Hence, parts of the population do no relate to institutions and rights at all. This constitutes a democratic deficit. Altogether, the assessment of the four elements of capacity demonstrates that actors’ capacity varies between the cases. However, in all 70 three cases, village chiefs have great capacity compared to other political actors. This fits with Törnquist’s (2002, 2004) argument that actors with the greatest capacity are the ones who combine single interests or communal loyalties with integration. Hence, in general the politicization and mobilization strategies actors employ are important explanations for their capacity to use and promote institutions and rights. Still, other explanations may be found as well, such as the conditioning factors that the following section presents.

7.3 Conditioning factors for actors’ capacity The framework of substantial democratization outlines that there are some conditioning factors that may influence and thus explain political actors’ capacity as well. These factors are actors’ political opportunity structure, their habitus, actors’ possession of social, cultural, economic, and their ability to transform these forms of capital into symbolic capital/legitimacy. This section will discuss the relevance of these factors for the cases. The discussion will reveal whether these aspects explain actors’ capacity, or whether the already listed strategies in the latter section fully explain actors’ capacity. Starting with the first factor, actors’ political opportunity structure, in Kontela and Tambaga, not all actors has equal political opportunities. This factor thus contributes to explain the variation in actors’ capacity. For instance, women and youth have restricted political opportunities in both cases. The population does not elect them as councilors, and they may not become village chiefs. However, their political opportunities are not so restricted in political parties. Here, young people are present in both places, while only in Tambaga, women participate in political parties. Women’s political opportunities are thus greater in Tambaga than in Kontela. The factor of political opportunities is fruitful for explaining this difference between women’s capacity in Kontela and Tambaga. Concerning the village chiefs, these have great political opportunities, because they condition the political opportunities of the populations by telling them which party to vote for in elections. Regarding the population in general, all castes have greater political opportunities in Kontela, because two ex-slaves have seats on the council there, while only nobles are councilors in Tambaga. Altogether, the examples discussed until now 71 demonstrate that political opportunity structure constitutes a fruitful explanation for actors’ capacity, as well as the variation between different actors’ capacity within the two cases. Further, it illuminates differences in the same actors’ capacity in the two cases, such as the varying opportunity of caste groups to get seats on the council in Kontela and Tambaga. In Mopti, clientelistic relations between founding families, the village chief and rich artisans on one hand, and small traders and poor on the other are important in conditioning actors’ capacity. Season work on the rice markets by the poor constitutes the base of these clientelistic relations (Bouju 2000:148). These relations are transferred to politics in elections (ibid.). The two–way dynamic of these relations provides benefits for both parties. In politics, the clientelistic relations between these groups change into neo-patrimonial relations, representing both hinders and opportunities to the clients in the relations. On one hand, they have the opportunity to change patron and thus influence politics indirectly. On the other, being part of such relations conditions their position, and they do not have political opportunities to engage in politics on their own outside these relations. Hence, a strategy of politicization and mobilization lies behind these relations. Still, the factor of political opportunities also contributes to explain the actors’ capacity, because the actors start out with different political opportunities. Consequently, together with actors’ politicization and mobilization strategies, the political opportunities explain actors’ capacity. One example that demonstrates this is that the poor do not have any political opportunities before engaging in clientelistic relations with the village chiefs and the founding families. The village chiefs and the founding families, however, have political opportunities before they engage in clientelistic relations. Still, their opportunities to influence the council are greater if they mobilize the poor in elections. Hence, they combine their already existing political opportunities with a clientelistic strategy of mobilization. Further, employees in the state administration have political opportunities that derive from their access to state resources they can use to build up clientelistic relations. As to the councilors, they have political opportunities, because of their position as councilors, and further their power over allocation and sales of property. Concerning political 72 parties, political opportunities in these are open for all, and the parties’ capacity lies in nominating candidates for elections. However, the village chief and founding families are the ones that mobilize the electorate, so the councilors do not have great independent political opportunities in elections (see sect 7.1.4). Mopti’s women have some political opportunities, because there are three women in the municipal council. This opportunity is greater than in the other two cases where no women are represented, and political opportunities are therefore a fruitful explanation for the difference in women’s capacity between the three cases. In general, political opportunities constitute an explanatory factor for most actors’ capacity in Mopti. However, in explaining actor’s capacity, this factor is less important than the strategies of politicization and mobilization that actors employ (see sect.7.2). Even if members of political parties enjoy some political opportunities outside the neo-patrimonial relations they engage in, such as by being members of parties that can nominate candidates in elections, neo-patrimonial relations constitute the basis for their opportunities (Bouju 2000). This implies that their strategies of mobilization and politicization elucidate their capacity, as well as the political opportunities they have as part of neo-patrimonial relations. Regarding the capacity of Mopti’s village chief and founding families, political opportunities do not explain the aspects lying behind that constitute these actors’ opportunities: gender, age, and the caste system. This applies to Tambaga’s and Kontela’s village chiefs as well. However, the second conditioning factor habitus, i.e. social norms/predispositions actors have, deals with these aspects and their importance for actors’ capacity. Further, habitus related to gender, age and the status one gets in the caste system structure all other actors’ political participation and representation as well, as described in previous section of this chapter. Still, the importance of habitus varies between Tambaga and Kontela. Since women in Tambaga participate in political parties, there are fewer social norms on women’s political participation compared to Kontela, and habitus is an explanation for the variation between Kontela and Tambaga concerning women’s participation in political parties. Further, the representation of the two ex-slaves on Kontela’s council 73 can be explained as that there are other norms than the one that only nobles are eligible at play in Kontela’s elections. However, habitus does not explain all actors’ capacity in the two cases. This is because if people act on the basis of habitus, politicians would not need to build up clientelistic relations with the electorate during elections, as they do. Consequently, habitus does not fully explain politicians’ capacity in the two cases. Still, habitus explains why Tambaga and Kontela’s councilors are all middle-aged men, and in Tambaga and Kontela only nobles are councilors (except 2 ex-slaves in Kontela). As to village chiefs’ capacity in Kontela and Tambaga, social norms/habitus in the caste system state that they are to be the chiefs, and these norms of the caste system is thus the foundation of their capacity. Additionally, the habitus in the sense of their position coming from their age and gender also plays a role, because the village elects the oldest man in the founder family to be the chief. Habitus represented by gender, age and caste therefore excludes women and youth from this position as well as in political representation, which explains their limited capacity in the two cases. Moreover, habitus is crucial in explaining why the two cases’ state representatives have no capacity, because social norms claim that only people from the municipality are to be political actors and legitimate power holders. Hence, habitus excludes the state representatives, since they are not from the municipality. Turning to Mopti, the ascribed status of village chiefs and founding families as Mopti’s founders constitute the habitus that condition these actors’ capacity. Age and gender are also part of the habitus on which their power/capacity is founded. However, even more than in Kontela and Tambaga, this habitus is not a sufficient explanation for their capacity to politicize and mobilize. Rather the neo-patrimonial relations they build up explain their capacity. This is also because the caste system is less important in Mopti than in Kontela and Tambaga. The only exception where caste matters in Mopti is in determining who belongs to the founding families and who is eligible for the village chief position, because the founding families are among the nobles in Mopti, and the village chief is elected among members of this family. A third external factor that may explain actors’ capacity is to possess various forms of 74 capital, such as social, economic, cultural and transform one or more of these forms into symbolic capital/legitimacy (Stokke 2002). All Kontela and Tambaga’s councilors have some symbolic capital, based on their cultural and social capital as the superior caste. However, the councilors lack economic capital. This deficiency explains why the municipal council does not play an important role in local politics in the two cases. Regarding village chiefs, they have cultural and social capital, due to the emphasis on customs and the caste system in politics. They also have economic capital through managing property distribution. They convert social, cultural, and economic capital into symbolic capital. Their strong position in the villages shows their symbolic capital/legitimacy in the population. Concerning political parties, they have economic capital in Tambaga and Kontela, as well as in Mopti, through funds that they may distribute to buy votes and alter the council’s power balance, making people change parties, which happened. As to the overall population of the three cases, it does not have any capital. Further, civil society has social capital in all three cases, because it constitutes an important economic support base for the population. Mopti’s chiefs over natural resources have economic and social capital, since they coordinate resource exploitation and organize work in different natural resources. Moreover, in Mopti, only the village chief and founding families have social, economic, and cultural capital. The status as Mopti’s founders is the foundation of the social and cultural capital these actors have. However, in order to maintain this capital, as well as their economic capital, the village chief and the founding families have to supplement their capital with clientelistic and neo-patrimonial relations. If they do not build such relations, their capital is not worth anything in politics. Concerning the councilors, their economic capital depends on their status as managers of sales of property in Mopti. The councilors’ economic capital through sales of property makes them necessary to collaborate with for the village chief and founding families. The village chief and founding families have legitimacy in the population, which the councilors need to stay on council. There is thus a mutual dependency in this relation between the councilors and the founding families/the village chief. This mutual dependency 75 explains the difficulties councilors encounter when they try to implement decisions without the consent of the village chief and the founding families (Bouju 2000:156; state employee 11.11.03). Further, the importance of property distribution explains the councilors’ great capacity in Mopti compared to the other two cases, where the village chiefs distribute property. As to Mopti’s state administration, it is the only out of the three municipalities’ state administrations which enjoys symbolic capital. Its symbolic capital depends on its economic capital, property, which it uses to engage in patronage practices with founding families and the village chief (Bouju 2000:147). All in all, despite the overall significance of the three conditioning factors in elucidating actors’ capacity, the politicization and mobilization strategies are also important explanations for actor’s capacity. Hence, a combination between the conditioning factors and the capacity to link activities, politicize interests, and mobilize people explain actors’ capacity.

7.4 Conclusion The three cases have the same outcome regarding the assessment of substantial democratization: some elements of substantial democratization have materialized in all three cases. Despite the same outcome, the cases differ on the extent to which the elements assessed are fulfilled. These elements are an elected municipal council and an administration, civil and political rights are mostly respected, a civil society exists in all three cases, and several actors have the capacity to use institutions and rights. Mopti’s institutions and rights are overall of better quality and scope than in the other cases. Kontela’s institutions and rights are of the poorest quality and most narrow scope. One explanation to this difference is Mopti’s long experience of local government and strong state presence, which makes it easier for its government to fulfill its tasks. The two rural cases have suffered from lack of state presence, and few available services. A third explanation is that many of the councilors in rural municipalities also lack knowledge of their tasks because they have received no information from the state. A fourth explanation, which counts for all three cases, is the constant deprivation of resources, because of little state transfers and lack of tax revenue. Yet an explanation lies in the disparity of actors’ capacity to link activities 76 in different parts of the political terrain, politicizing issues, mobilizing people, and using institutions and rights. For instance, Mopti’s councilors have more capacity than in the other two cases, because they enjoy economic capital, which makes them important to cooperate with for the village chief, ward chiefs, and founding families. In the other two cases, the village chiefs distribute property. Further, the village chiefs have the greatest capacity in all three cases, because they are able to link activities in several parts of the political terrain, mobilize the population and influence local government. They also have symbolic capital. A fifth explanation is that most actors politicize single issues and only use institutions and rights, and thus do not promote substantial democratization. Altogether, the explanations listed here demonstrate some of the main causes for why decentralization has not yet materialized in full substantial democratization in the cases. For substantial democratization to be realized these issues have to be dealt with. The most important challenge for the three cases, is however to restructure local power relations, in order for equal participation and representation of all social groups, age groups and both sexes to materialize. This implies that one has to find a way to undermine the sanction power of the village chief and founding families in Mopti, and the chiefs’ impact on decisions of the councils in Tambaga and Kontela. To do this is crucial, because their presence hinders development and democratization. However, it does not imply that one should abolish the chiefs, because they also have an important function as representatives for their villages towards the council. Rather, one has to find a way to preserve their representative role, while at the same time undermine their negative influence on political participation and representation. In addition, for full development of substantial democratization, the municipal council and the state has to build up their legitimacy through democratic decision-making instead of through single-issue politics and neo- patrimonial relations, as has been the case from 1999 until the end of 2003. 77

8. General explanations and theoretical arguments

The two analytical frameworks employed in the previous chapter contain general explanations for why decentralization only has contributed to realize a few elements of substantial democratization in the three cases. In this chapter, I will discuss these explanations. However, in order to explain the cases’ outcome in detail, as well as the variation between the cases, there is a need for more specific theoretical explanations developed from the African political context. The thesis includes three theoretical approaches that have specific theoretical explanations for the result of the assessment of the cases. This chapter will elaborate on them. In the following chapter, I will employ and test the theories’ arguments/explanations on the cases. The theories were chosen, because they represent different theoretical approaches to the linkage between decentralization and democratization. Besides, these theories are fruitful to use, because they are recent contributions to the theoretical debate on this linkage and the overall theoretical debate on African state and politics (see ch. 2).

8.1 General explanations for the cases’ outcome This section will use the explanations that Törnquist (2004) and Crook & Manor (1998) bring forward to explain the cases’ outcome. One of the factors Crook and Manor outline is that the national and local political and social contexts matter. The variation between the three cases indicates the context’s importance in them. For instance, Mopti’s long experience of local government is a significant explanation for the local government’s higher accountability, compared to the other two cases. Further, nobles’ domination over the municipal council in Tambaga and Kontela demonstrate that local elites have captured local government. The strong position of the village chief and founding families in Mopti vis à vis the council also shows elite capture to some extent. Altogether, these three examples illustrate the crucial role of the local political context in determining the result of decentralization. Concerning social context, usage of the caste system is an example of the significance of the social context in politics, which further explains the lack of full democratization in all three cases. Another essential explanation for the authors is the presence of 78 competitive political parties. In all three cases, there are many political parties, but they are highly unstable and not very competitive, because most members have joined them in order to gain access to resources. Hence, members easily change parties, which make it difficult to build up a sustainable local party organization (see ch 6 and 7). Furthermore, according to the authors, a widely distributed free press is vital for democratization. The press is free in Mali (URL:www.rsf.fr), but in Tambaga and Kontela no press exists, providing one explication for lack of accountability here. In Mopti, there is a widely distributed press, used to spread information from the council. Due to this press, people know more about the municipal affairs in Mopti than in the other cases, which contributes to explain the better performance regarding accountability of Mopti’s local government. Also, the factor of having a professional national civil service constitute an explanation to the cases’ outcome regarding substantial democratization, because such a service is not present in Kontela and Tambaga, since the state administration there has low quality. Mopti’s state administration has mediocre quality, assisting the local government in many matters, which elucidates why Mopti’s local government is more accountable than the ones of the other two cases. Turning to Törnquist’s explanations (2002, 2004), they overlap with the one on the context’s importance by Crook and Manor, and help specify the significance of the cases’ context for the findings in chapter seven. Törnquist’s first explanation is that substantial democratization will seldom follow from elite-based decentralization, because such reforms do not change power balances nor do they open for popular political participation. Although Mali’s decentralization reform aimed at involving the population and in some localities, it did, it has not happened in Kontela and Tambaga. Consequently, people were not informed about decentralization. Hence, they lack capacity to use institutions and rights, which results in elite capture of local government, as outlined above. Törnquist’s second explanation is the incapacity to link activities in different parts of the political terrain. Most political actors can link activities in various parts. The village chiefs are the ones with highest capacity to link activities, because they are in 79 three parts of the terrain (see sect 7.2). Their capacity, as well as the population’s incapacity makes this factor able to illustrate why full democratization has not materialized in the cases. The third explanation is the tendency to politicize on single issues and mobilize people via clientelism. In the three cases, the village chiefs and founding families have more capacity to mobilize the population than local government. These forces and some of the councilors follow single-issue politics, which obstructs work in the three councils, because to profit from engaging in politics is the motivation factor for most of the councilors. Törnquist’s fourth explanatory factor of substantial democratization’s failure is non-existence of public spaces where one can mobilize political issues. Although some public spaces exist in the cases, such as public meetings, political meetings in parties and various associations, when they involve political matters, they usually exclude women and youths from participating, although the exclusion varies between the cases. While these spaces only mobilize middle-aged men in Kontela, some women also participate in Tambaga. In Mopti, public spaces are more open, and all groups may partake. However, having democratic public spaces have not resulted in council representation for all social groups, both sexes, and different age groups in any of the cases. Altogether, Törnquist, as well as Crook and Manor provide some relevant general explanations for the result of substantial democratization in the cases, but do not explain all differences between them.

8.2 The three specific theoretical approaches 8.2.1 Political instrumentalization of disorder? One of the three theories discussed in this thesis is the one of Chabal and Daloz (1999) on African state and politics: the political instrumentalization of disorder. This theory argues that colonial powers never institutionalized the colonial state. Current African states are thus not “hybrid” states with colonial and pre-colonial features, as Bayart (1996) and various neo-patrimonial perspectives claim (Chabal and Daloz, 1999:11). Rather, African states have reshaped the traits from pre-colonial, colonial, and post-colonial times into distinct African states. A political disorder of instrumentalization characterizes current African states. This disorder has several 80 traits, which people instrumentalize. Consequently, all people’s actions are “informalized and personalized”(ibid.:1). The traits are based on African culture, and they are overlapping (ibid.:147, 149). The two central features of the disorder are the communal belonging and reciprocal and vertical relations. These two features make other traits of the disorder appear. People instrumentalize all these traits of the disorder in politics, resulting in the political disorder of instrumentalization. The communal belonging implies that people identify themselves as being part of a community, i.e. village, family, ethnic group, caste, among others (ibid.:156). They found their identity, rationality, and actions on this belonging. Identity is thus more inclusive and extensive than in Western countries, because it embraces several belongings people have. These belongings are instrumentalized in politics. These two traits thus influence political participation and representation by all people in a state. Identity’s extensiveness makes politics encompass several spheres, such as public, private, religious, and family spheres. All actors instrumentalize these spheres when profitable (ibid.159). Hence, the instrumentalization of communal belonging, as well as reciprocal and vertical relations leads to another trait of the theory: non-separation between public and private spheres. Moreover, civil society does not exist, because of non-separation of public and private spheres where people instrumentalize their position in both spheres. This instrumentalization also accounts for civil society, and no associations are thus independent from the state. Hence, they do not form a civil society (ibid.:21). Due to this non-separation, public employees will seek to profit on their positions, exploiting state resources (ibid.). They will also instrumentalize customary actors, such as village chiefs when they gain from it (ibid.). Concerning the feature of reciprocal and vertical relations these are instrumentalized in politics, because such relations between politicians and electorate determine both parties’ political actions (ibid.:158). Building such relations means extending the communal belonging, because these relations create such a belonging where it does not exist. There is, though, an underlying premise of patron- client relations: If the patron fails to fulfill his part of the relationship, the client may change to another patron (ibid.). Moreover, such relations may take several forms, 81 but the authors concentrate on clientelistic (personalized) and neo-patrimonial (non- personalized) forms (ibid.). Another factor of the disorder, which derives from the vital role of reciprocal and vertical relations in politics, is that political success, accountability, and legitimacy is measured in achievement of resources, such as gifts and property. Politicians instrumentalize the material goods they have by showing them off and distribute them to their clientele. This way they obtain accountability and legitimacy. Hence, short-term goals characterize politics. Such goals are another trait of the theory. Consequently, ideological arguments are not important in politics, and political parties do not stay in opposition if they do not profit from it in terms of resources gained (ibid.:55). Another trait of the instrumentalization of disorder is recycling of political elites. Even if new political actors appear, they will instrumentalize their actions as the old ones do, because they profit from doing so. Altogether, political actors instrumentalize the features listed here, which results in the current political disorder of African states, i.e. the Africanization of democracy (ibid.:146). Regarding the cases’ outcome, this theory may help explain the weak position of political parties, the strong role of some political actors compared to others, as village chiefs, as well as why neo-patrimonial relations are important in Mopti’s politics.

8.2.2 A bifurcated state? Mamdani’s (1996) theory of the bifurcated state concentrates on differences between African urban and rural local governments. These differences, as well as the element of lacking balance between key elements, described later on, explain why decentralization does not lead to democratization in Africa. Mamdani criticizes all perspectives emphasizing on patrimonialism for downplaying colonial rule’s impact on the character of current African states (ibid.:13). This criticism hits the theory of Chabal and Daloz, as well as other theories presented in chapter two. Mamdani argues that the colonial racial domination system has continued in all African local governments. This racial system was dual, separating between citizens and subjects in the states. Urban areas enjoyed civilized direct rule consisting of civil laws and a modern state administration and government for the “civilized population” (ibid.) 82

(mainly colonizers). In rural areas, there was indirect rule by customary chiefs, whom colonial powers had appointed (ibid.:15). These chiefs were decentralized despots, because they had monopoly of power in rural areas, which the state gave them. Being state representatives at the local level, decentralized despots ensured the state’s control over rural areas (ibid.:300). In order for these chiefs to be loyal against the state, the colonial powers appointed other chiefs than the pre-colonial ones. Hence, they transformed the customary chief institution into an all-encompassing ethnic power who ruled by single customary law in rural areas (ibid.:122). After independence, Mamdani argues that this bifurcated state has persisted, either as decentralized despotism in rural areas, already described, or as centralized despotism for the entire country. Centralized despotism is a system where local party cadres replace customary chiefs and enforce the central state’s repression at the local level. This bifurcated state with a rural-urban divide makes civil society bifurcated as well; being civilized in urban areas, i.e. independent, and tribalized in rural ones, i.e. influenced by customary chiefs. Unless one abolishes the rural-urban divide, and thus the bifurcated state as well, decentralization will not lead to democratization. Mamdani argues that in order for decentralization to result in democratization, one also needs balance between decentralization and centralization, representation and participation, and autonomy and alliance. Without a balance between these elements, decentralization will only worsen the rural-urban divide. Thus, democratization will not succeed (ibid.:298). In this thesis, Mamdani’s theory may explain the differences found between the rural and urban cases. It may also help explicate the significance of village chiefs and customary law in the rural cases.

8.2.3 Local politics- a complex game The third theoretical approach is the one by Bierschenk and Sardan (1998). This theory argues that fragmentation characterizes local politics, because many institutions, both informal and formal, customary and “modern” ones exist. These institutions constitute different local power arenas, with diverse rules, legitimacy, and actors. All these arenas have a fluid character. The state is thus only of several local power arenas, and does not have regulatory power locally (ibid.:23). Hence, the 83 situation is far from one of decentralized despotism, where the state has penetrated the local level (ibid.:23). Because of the existence of many power arenas and fragmentation of power, local politics depends on the actors present and their ability to negotiate and capture power (ibid.). One of the arenas that contribute to fragmentation is development brokers, who are educated people and often leaders of local associations or NGOs, or assisting these (Bierschenk et al. 2000). They fragment local politics, because they replace the state in service delivery, and thus help de-legitimize (Bierschenk and Sardan, 1998:44). Even if the theory focuses on fragmentation of state and not local government, the thesis discusses fragmentation of local government instead of the state, since the cases’ local governments are partly responsible for public services. Further, the theory argues that decision-making is not transparent, because it has an informal and oral character. For democratization to result from decentralization, the state must gain legitimacy in the population and control the actions of other local institutions. Also local government has to obtain institutional and popular accountability. As to the cases, the theoretical arguments may explain the low accountability of the state and local government, among other elements.

8.3 Conclusion The general explanations of Törnquist and Crook and Manor help explain the cases’ outcome. Among these explanations are deficiency of resources, no presence of competitive political parties, and the domination of single issues in politics and many more. However, even if these and other explanations the frameworks present focus on the importance of the local context for the result of democratic decentralization, they are still general explanations. The thesis therefore includes contextual and specific theoretical arguments derived from three theories, outlined in this chapter as well. The next chapter will show whether these arguments explain why decentralization has only contributed to the realization of some elements of substantial democratization in the three cases. 84

9. Assessing the theoretical arguments In this chapter, I will employ the specific theoretical arguments introduced in the previous chapter to explain why decentralization has contributed to the realization of only some elements of substantial democratization in the cases. The theoretical arguments provide diverse interpretations of this result concerning substantial democratization, and they may thus illuminate significant aspects of the cases. Applying them allows for critical examination of their relevance to the cases as well. This chapter also compares the situations before and after establishment of local government, by undertaking a within-case analysis using most similar systems research design. This analysis will contribute to increase knowledge of the decentralization process and its result in the cases. The three sections that apply and assess the relevance of the theoretical approaches start with questions that illuminate the elements to be examined in each approach.

9.1 A bifurcated state? In this section, I will discuss and assess Mamdani’s theoretical arguments on the cases. I will do this by answering the following questions:

-Do customary chiefs wield authority in Kontela and Tambaga through representing the state? Are there any customary chiefs in Mopti? Do municipal council and village chiefs apply customary law in the three cases? Are people treated as citizens in Mopti and as subjects in Kontela and Tambaga? Do customary chiefs condition civil society in Kontela and Tambaga? Is civil society in Mopti free from the influence of customary chiefs? Does decentralized despotism exist in Kontela and Tambaga?

-Is there a balance between decentralization and centralization, between autonomy and alliance, and between participation and representation?

Mamdani argues that African states are characterized by either an overall system of centralized despotism or a system of decentralized despotism in rural local governments. Centralized despotism implies that the central state governs through local party cadres, often within a single party system. Such a system is more 85 descriptive of Mali’s situation before implementation of national democratic rule in 1992 than after decentralization (see sect.9.4). Since this section only deals with the period after decentralization, I therefore assess whether the rural cases are marked by decentralized despotism. In section 9.4, I discuss whether the cases are in a state of centralized despotism, because I examine the cases before decentralization and even before establishment of national democracy in 1992. Turning to the first question of this section- whether customary chiefs wield authority in the two rural cases- they do. There are customary chiefs with political power in Mopti, i.e. the village/ward chiefs. The difference between the current village/ward chief institution in Mali and the decentralized despots (Native Authorities) Mamdani describes is the state’s influence over the chiefs. While Mamdani argues that the state appoints these chiefs and influences them, the Malian state’s approval in the election of village chiefs is only a formality (Tag 1994:87). Males in villages/wards elect their chiefs. Hence, the state does not have an impact on this election. Instead, age and lineage are the principles that determine eligibility in this election (Beridogo 2002). However, the importance of these eligibility criteria differs between the two rural cases and the urban one. In Kontela and Tambaga, the criteria of village chief election are lineage and age. Lineage means that one has to be of the founder family, and age that the oldest living son or brother of the deceased chief is elected as chief (République du Mali 2002; Tag 1994:86). In Mopti, age and lineage always apply as criteria when electing the village chief, but not when electing ward chiefs. Wards chiefs may come from any family and they may be young. This difference implies that changes have happened in Mopti’s ward chief institution (state employee 13.11.03 [interview]), which Mamdani cannot explain, because he does not acknowledge that these chiefs exist in urban areas. According to Mamdani, customary powers would only wield influence in rural areas, so Mopti deviates from his theory, and he does not explain Mopti’s contradictory situation. Moreover, the village/ward chiefs wield influence on local government in all three cases, but they do not monopolize their power. The elected local government has undermined their monopoly, because these governments also have an impact on politics in the three 86 cases. Consequently, the cases’ village/ward chiefs are not decentralized despots, because their power differs from the absolute power enjoyed by the customary chiefs that Mamdani describes. Nonetheless, the way the chiefs’ influence has changed since decentralization varies between the cases. Both in Kontela and Tambaga, the village chiefs have strong positions in villages, because the population regards them as authorities expected to wield power. In addition, the chiefs have an impact on the council’s decisions. However, they wield more influence over Tambaga’s council than Kontela’s, because Tambaga’s council frequently consults the chiefs, while Kontela’s council does not. Still, less consultation and influence in Kontela does not imply that village chiefs wield less influence in local politics. The reason is rather that Kontela’s municipal council takes few decisions, due to that the council lacks resources, is therefore not very active, and thus does not have much impact on people’s lives. Regarding Mopti’s village chief, he influences the population in elections, as well as the municipal council, but neo-patrimonial relations between him and the founding families on one side and municipal councilors and other politicians on the other determine his power (Bouju 2000). His status as customary power does therefore not automatically make him a powerful actor, as in Kontela and Tambaga. Hence, the basis for village chief’s authority differs between the two rural cases and the urban one; founded on neo-patrimonial and clientelistic relations in Mopti, and based on the caste system in the other two cases. The theory fails to explain this divergence, because it does not acknowledge that customary chiefs wield power in urban areas. Mamdani also argues that one ethnic group dominates the village chief institution in rural areas. This argument applies to all three cases, because their village chiefs are from one ethnic group that excludes members of other ethnic groups from becoming chiefs. Concerning whether Tambaga’s and Kontela’s populations are subjects, they indeed are, because village chiefs enjoy power over them and apply customary law. This theoretical argument thus contributes to explain the two cases’ outcome regarding substantial democratization. As to Mopti’s population, they are also subjects under the village chief and founding families, due 87 to their clientelistic relations with the latter actors (Bouju 2000:151). However, the application by Mopti’s judiciary of modern and customary laws ensures equal treatment before the law to some extent, and therefore Mopti’s population are sometimes citizens rather than subjects. Since the theory argues that people are treated as subjects only in rural areas, and not in urban ones, it cannot describe Mopti’s ambiguous situation. Moreover, according to Mamdani, rural and urban civil societies differ from each other, because rural civil society is tribalized, implying that village chiefs and other customary chiefs have an impact on it, while urban civil society is civilized, i.e., free from village chiefs’ influence. Mopti’s urban civil society is tribalized and civilized at the same time, and the case thus partly deviates from the theory on this point. Mopti’s civil society is tribalized, because in the natural resource associations, customary leaders condition leadership. Concurrently, civil society is civilized, because several independent organizations without customary leadership exist. Turning to Kontela and Tambaga, economic activities organize associations. It has, however not been possible to document systematically whether these associations are tribalized, i.e., whether they are under village chiefs’ control, because these associations are so numerous and there is no literature on them. However, in most of the associations in the region to which these cases belong village chiefs have an impact on decision-making and leadership election (Davies 2000). In each of the cases, for example, the leader of one of the women’s associations is one of the village chief’s wives (woman 17.10.03 [interview], woman 29.10.03 [interview]). Due to the village chiefs’ powerful position, they are likely to have a say in most associations. The rural cases thus have tribalized civil societies, and therefore fit Mamdani’s theoretical argument. In general, Mamdani argues that one needs to abolish the divide between urban and rural local governments in order for democratization to materialize. As to the cases, a divide only exist to some extent, since a village chief and ward chiefs exist, as well as both civilized and tribalized civil society, and customary law and modern law is present in the urban case Mopti. Consequently, a there are only some elements of a bifurcated state in Mali taking into 88 account these three cases. The argument of a bifurcated state can therefore not provide a thorough explanation to the cases’ outcome, and in particular not to Mopti’s contradictory situation. However, Mamdani has yet another argument concerning decentralization and democratization that has to be tested on the cases. This is that democratization requires balance between the elements in each of the following pairs: centralization and decentralization, autonomy and alliance, and participation and representation. Concerning centralization and decentralization, there is more centralization than decentralization in all three cases, because the state has not devolved all its responsibilities to local government (se ch 7). Therefore, local governments still rely on state transfers. As to the balance between autonomy and alliance, the state funds most of Kontela’s and Tambaga’s services and salaries for its employees, and thus the situation of these cases is predominantly one of alliance. Mopti’s council has greater autonomy than the other two, because its councilors are more experienced and it has independent revenue from the sales of property. Nevertheless, due to its dependence on state transfers, Mopti is also in a situation of alliance. Hence, none of the cases has attained balance between autonomy and alliance, nor between representation and participation. In Kontela, though, all caste groups participate in party politics, only two groups have seats on the municipal council. Tambaga’s situation is even worse regarding balance, since only male nobles are on the council, despite all groups and both sexes participate in party politics. In Mopti, political participation in parties is fully open to all groups and both sexes. However, the village chief and founding families influence council representation. Consequently, balance between participation and representation does not exist in any of the cases. In general, the balance needed between decentralization and centralization, autonomy and alliance, and participation and representation to achieve democratization according to the theory is not present in any of the cases. Therefore, this argument of the theory helps explaining why the cases do not fulfill all elements of substantial democratization. Altogether, the arguments of the theory assessed here partially give an account of why only a few elements of substantial democratization have materialized in the 89 cases. However, it cannot explicate the contradictory situation in Mopti, where both customary law and modern law is applied, a tribalized and a civilized civil society, as well as customary chiefs are present. Hence, the theory does not provide a comprehensive explanation of the cases’ respective outcome regarding substantial democratization.

9.2 Political instrumentalization of disorder? In this section, I will apply the theory of Chabal and Daloz (1999) to the cases by asking and answering the following questions:

- Is the traditional sphere instrumentalized in politics? -Does communal belonging exist and is it instrumentalized in political participation and representation in the cases? -Are reciprocal, asymmetrical relationships between politicians and voters, as well as between other political actors present? Are these instrumentalized? - Do actors instrumentalize material goods and services to obtain political legitimacy and accountability? -Is there non-separation between public and private spheres? If so, do actors instrumentalize this non-separation? -Are political actors instrumentalized in the sense of being recycled? -Do people instrumentalize civil society?

Regarding the first question of instrumentalization of the traditional sphere, I discuss only village/ward chiefs, because these are the most important actors in the traditional sphere in the cases. The discussion would be too broad if I were to include all actors and aspects that are included in the traditional sphere. The theory argues that all people instrumentalize the traditional sphere, i.e. the village/ward chiefs, when they can profit from doing so in politics. In the three cases, the councilors and politicians are not able to instrumentalize these chiefs, because the chiefs have strong political positions and influence local government and politics. They have an impact on the three local governments, because the governments need the approval of village chiefs in decision-making. If the chiefs are not asked to approve decisions, then they are 90 likely to hinder local governments’ implementation of decisions (see ch. 7). The theory cannot give an account of the underlying causes for why political actors are not able to instrumentalize village/ward chiefs. This constitutes a weakness in the theory’s explanatory value in the three cases. Still, the theory can explain why the village chiefs are important actors in the three cases, by turning to another of the theory’s features: communal belonging. The communal belonging of the village/ward chiefs includes their belonging to a certain caste and to a village. The village/ward chiefs instrumentalize their standing as being of the highest caste (the nobles) in the social hierarchy in a village, as well as being the political and administrative head of a village/ward. The chiefs instrumentalize these positions in order to gain political power in the municipality. However, the chiefs instrumentalize these two positions in various ways in the three cases. In Kontela and Tambaga, village chiefs instrumentalize their status/belonging in the villages to influence decision-making in the local governments. They do this through meetings with the governments or through municipal councilors outside meetings. The chiefs also have an impact on implementation of decisions, as previously mentioned. The difference between these two cases and Mopti is illustrated later on in this section. In addition to explaining the strong position of the village chiefs, the caste system as a communal belonging also determine political participation and representation on local government in Kontela and Tambaga. The significance of the caste system is demonstrated by the fact that only people of the upper caste, the nobles, participates and is represented on Tambaga’s council. This caste group thus instrumentalize their position as being of the upper caste in politics. With the exception of two ex-slaves, all Kontela’s councilors are nobles as well. Nonetheless, in elections, political instrumentalization of the caste system does not suffice to mobilize the electorate. It has to be combined with political instrumentalization in the form of gift giving. Gifts are used to build reciprocal and vertical neo-patrimonial relations between candidates and the electorate. The presence of two ex-slaves on Kontela’s council also shows that instrumentalization of the caste system is not the only factor determining political participation and 91 representation. Further, regarding participation in political parties, political instrumentalization of the caste system does not condition such participation, because all castes participate in political parties in the two cases. Hence, instrumentalization of the communal belonging as in belonging to a caste does not explain all political participation and representation in the two cases, only participation and representation on the municipal council. If one looks at other elements that are included in people’s communal belonging, such as belonging to a family, an ethnic group and a village, some of these belongings help illustrate people’s voting in the two cases. For instance, there is political instrumentalization of family belonging in the two cases, because some people vote for family members who are candidates in elections. These candidates tell their family to do so. Likewise, several people vote for candidates/parties which family members instruct them to vote for. For instance, husbands may tell their wives on which party they are to vote. Even if the family belonging is thus present, the reason for why people vote according to their family belonging is owing as much to their lack of knowledge of politics as to people instrumentalizing their family belonging or letting candidates instrumentalize people’s belonging to make them vote for their party. This implies that the family belonging does not explain all people’s voting in the two cases. As to the instrumentalization of village belonging, village chiefs decide which party the population is to vote for in some Tambaga villages (municipal employee 28.10.03 [interview]). Consequently, they both instrumentalize their status as power brokers in the village, as well as people’s belonging to the village. In all, the examples outlined above on instrumentalization of the caste, village, and family belonging demonstrate that communal belonging is present in politics in the two cases, and explains people’s participation and representation to a great extent. However, communal belonging does not suffice as a general explanation to all people’s political actions. This theoretical factor thus has to be supplemented by other traits of the disorder of instrumentalization that the theory outlines, discussed later on in this section. Turning to Mopti, the village chief and the founding families, and to some extent the ward chiefs, are able to instrumentalize their status as founders of 92

Mopti city in politics just like in the other two cases. By instrumentalizing this status, they use their identity as founders. This identity stems from the communal belonging they have to Mopti as a city. This status is also related to another feature that communal belonging includes, the caste system, because the status of the chiefs and the founding families derive from being of the upper caste of nobles, as well. Communal belonging in the form of the position these people have in the caste system thus has an impact on political participation and representation in Mopti. For instance, belonging to the caste of nobles and to the founders of Mopti explains the participation of founding family members in politics, because they participate and influence politics because of their belonging, i.e. their status as founders. Hence, these families instrumentalize this status in politics. Still, their position does not guarantee that they get seats on the municipal council or political influence in other ways. The status as founders and nobles does therefore not suffice as an explanation to the significant role of the village chief and the founding families in Mopti as it does in the other two cases. The founding family members that stand for elections have to engage in neo-patrimonial relations with voters in order to gain representation and influence politics. Further, the village chief, ward chiefs and founding families also have to engage in neo-patrimonial relations with the electorate and political parties in order to ensure their influence over political participation and representation in Mopti. Furthermore, these chiefs and families must maintain the neo-patrimonial relations after elections so as to be able to instrumentalize the relations they have built up. They instrumentalize the neo- patrimonial relations with the local government in sales of property among other things (see ch 7). All told, with the exception of people’s participation in parties, instrumentalization of the caste system, as well as of village and family belonging explain politics in Kontela and Tambaga, and to some extent in Mopti. Since these three belongings are included in the factor of communal belonging, the communal belonging is thus instrumentalized in all three cases. It is however less important in politics in Mopti than in the other two cases. The theory makes clear that such variation in the instrumentalization of communal belonging as well as in all its other 93 theoretical factors, may exist between countries, and thus also between these three cases. Therefore, the theory can explain the differences between the cases concerning the role of instrumentalization of communal belonging in politics. Nevertheless, the theory does not provide the underlying causes for why communal belonging is not important in all politics in the three cases. This reduces the theory’s explanatory value. Turning to discussing another of the theoretical traits of the political disorder of instrumentalization; political instrumentalization of reciprocal and vertical relations, it has already been stated in this section that this trait is present in the three cases. In Kontela and Tambaga, reciprocal and vertical neo-patrimonial relations exist between politicians and electorate during municipal election campaigns and these are instrumentalized by both parties. The politicians build these relations on gift giving and instrumentalize them by making the electorate vote for their party in elections. The electorate also instrumentalizes these relations, because it votes for the party from whom it receives gifts, according to several informants. The political instrumentalization of these relations is however short-term, because they are only present during municipal election campaigns and elections. Furthermore, in some instances, voters in Kontela turn these relations upside down, because voters instrumentalize politicians’ effort to create neo-patrimonial relations, instead of letting politicians instrumentalize their voting (state employee 11.10.03 [interview]). They do this by receiving gifts from all parties, promising to vote for them. The theory argues that such a situation appears because the patrons (politicians) do not fulfill their part of the relation, which makes clients entitled to break it. However, this explanation does not count for Kontela’s condition, because people exploit simultaneously several patrons, with whom the relations remain intact. Still, even if reciprocal and vertical relation cannot explain this situation in Kontela, the situation is an exception to the general picture of political participation there. Hence, in most instances, instrumentalization of reciprocal relations explicates political participation and representation in Kontela. The mayor’s election in Kontela demonstrates that instrumentalization of such relations exists. In this election, district level politicians in the ADEMA party 94 established neo-patrimonial relations with municipal councilors in order to make the ADEMA candidate win the election, which he did. He won the election because the district politicians were able to make some councilors of the opposing side change side by promising them money. Thus, both the district level politicians and the councilors who changed side instrumentalized the reciprocal and vertical relations that they had constructed between them. Regarding Mopti, instrumentalization of reciprocal and vertical relations determines peoples’ political participation and representation, because people vote according to these relations and thus instrumentalize them in voting (ibid.:152). They act as they do because their family, their ethnic group, or the caste that they belong to, that is their communal belonging, do not provide them with what they need to survive. Hence, they have to seek patrons outside this communal belonging, and engage in reciprocal and vertical clientelistic relations with the village chief, ward chiefs, and founding families. Since large parts of the population, not only the poor, rely on instrumentalization of these relations, politicians wanting to mobilize voters also have to maintain and instrumentalize them. The village chief, the ward chiefs and the founding families are patrons in these relations, while politicians depend on these patrons to mobilize voters (Massing 1996a). Once elected, municipal councilors have to serve these patrons, such as in the sales of property (see ch 7). These patrons then provide their clients (the population) with benefits that they have got from their relation with the councilors, such as property or other resources. Thus, all parties in these relations benefit from them. Even if politicians have to go through the village chief, ward chiefs, and founding families to mobilize the electorate in elections, they are still patrons in the neo-patrimonial relation that they enjoy with the electorate. This is because the politicians offer gifts or promises to the population in elections and give favors to the village chief and founding families in the sales of property (see ch. 7). The latter point shows that Mopti’s situation coincides with yet another of the theory’s factors, that politicians instrumentalize material goods and wealth they have in order to achieve accountability and political legitimacy towards the electorate. They do this by 95 distributing their goods to the electorate or by displaying their wealth to the electorate to demonstrate that they are successful. However, even if this factor is descriptive of how politicians achieve legitimacy and accountability in Mopti, it seems like there is a reduction in the municipal council regarding the importance of displaying wealth or distributing goods for the council’s accountability and legitimacy. Due to this reduction, the significance of neo-patrimonial relations also decreases, since these relations are often used in distributing goods. One example of the decrease of these two features’ importance is that the council does not always consult village chief, ward chiefs or founding families in decision-making, and takes decisions that oppose these latter actors’ views. For instance, during the decentralization reform, the council did not consult these actors (Massing 1996a; see ch 7). Instead, it decided on matters itself concerning decentralization, and it has done so on other issues as well after decentralization. The council thus attempts to take democratic decisions, trying to avoid using material goods and wealth, as well as the neo-patrimonial relations to show legitimacy and accountability. Instead, the council tries to seek legitimacy and accountability in the population by taking democratic decisions. By doing this, they undermine their neo- patrimonial relations, because they act independently of these relations. The theory argues that such a change in the council’s actions is not likely to take place in the near future in African states, since all people; both politicians and the electorate, profit more from instrumentalization of reciprocal and vertical relations than democratic decisions. The reduction in instrumentalization of goods and services as well as neo-patrimonial relations in Mopti’s council is thus a deviation from the theory, and the theory cannot elucidate the underlying causes for why this change takes place. Nevertheless, this is a minor reduction, because in most instances, instrumentalization of neo-patrimonial relations and distribution of goods and displaying wealth are the basis for the council’s accountability and legitimacy. Hence, in general, these factors explain Mopti’s situation. In Kontela and Tambaga, councilors try to gain legitimacy and accountability before elections by gift giving. They succeed, as previously described in this section. Although councilors create 96 short-term neo-patrimonial relations with such gift giving, they do not follow up on them after elections, and the relations break up. Hence, providing goods, services and giving gifts constitutes a means for political parties to obtain votes in elections. Gifts and services are however no foundation for the accountability and legitimacy of the two local governments in between elections. Instead, Tambaga’s government is somewhat accountable through making public services available for the population, while Kontela’s government is not accountable at all, because it provides few such services. The theory does therefore only explain Kontela’s and Tambaga’s circumstances regarding accountability and legitimacy in elections. The theory argues that if displaying goods and wealth are not important for accountability and legitimacy, it only implies that some of the other factors of the theory are likely to explicate how local government obtains accountability and legitimacy in the cases. For instance, local government may achieve accountability and legitimacy by instrumentalization of reciprocal and vertical relations or the communal belonging. However, as previously concluded, neither instrumentalization of the communal belonging, nor instrumentalization of reciprocal and vertical relations can explain why instrumentalization of material goods and wealth only illustrate the accountability and legitimacy of the two local governments in elections. It thus seems like the theory fails to explain the underlying grounds for the accountability and legitimacy of Tambaga’s local government in between elections, and the lack of accountability and legitimacy of Kontela’s government. However, it is still present in elections. Further, since this argument explains the basis for accountability and legitimacy of Mopti’s government in most instances, and in elections in the other two cases, the argument has overall good explanatory value for the cases. Moreover, according to the theory’s argument concerning recycling of new and old political actors, all actors are recycled. That they are recycled means that they instrumentalize their political actions. Even new political actors are recycled, if they do not contribute to political changes in the local government’s politics, but rather instrumentalize their political actions to gain instrumental benefits. Mopti’s new actors, such as municipal councilors and 97 municipal employees have changed the politics of the local government to some extent. The local government’s many improvements in service delivery over the last years reveal this change. Hence, some of the new actors are not recycled in Mopti. Tambaga is in the same situation, having both old and new actors, but some new ones contribute to improvements in services (see sect.7.1.3), and it thus seems like they are not recycled. Kontela’s political actors, though, are recycled, because old actors, such as the previous mayor, undermine the local government’s new actors. The previous mayor was an old political actor from the time of the one-party state, and had a strong position in the municipality in general, as well as in the municipal council (Amundsen 04.02.04 [interview]). He was thus able to hinder the political initiatives from new councilors from being realized, because the majority of the councilors voted according to his will (ibid.). Hence, recycling is present in Kontela, but only to some extent in Mopti and Tambaga. Recycling of political elites thus provides an explanation to Kontela’s result regarding substantial democratization, while the factor only partly exists in Mopti and Tambaga. As to the theory’s argument about non-separation between public and private spheres, this argument implies that people instrumentalize their position in public office to enrich themselves or other people they have a relation to. This non-separation is descriptive of the work of the Malian central state administration because state employees engage in clientelistic practices (Lange 1999). Turning to the cases, in Mopti it seems like property management exemplifies this non-separation. In Kontela, the local government’s mediation of conflicts and abuse of tax money shows that non- separation is present there (municipal employee 06.10.03 [interview]; anonymous 07.09.04 [e-mail]). In Tambaga, non-separation cannot be documented. Hence, the theory fits with Mopti and Kontela. Further, the cases differ concerning the theoretical aspect of instrumentalization of civil society. According to the theory’s definition of civil society (see ch 8), all primordial associations, i.e. those based on communal belonging in any form, are excluded from civil society. With the exception of the emigrants’ associations, all other associations in Kontela are based on village 98 belonging, and are not part of civil society. The emigrants’ associations do not form a civil society either, because they do not comprise an independent civil society, since some of their local leaders in Kontela are also councilors, and there is thus no separation between the spheres (councilor 16.10.03 [interview]). In Tambaga, a civil society does not exist, because all associations are primordial ones. In Mopti, civil society is present, because of the existence of several non-primordial organizations. It is not possible to document that all these organizations are independent from reciprocal relations with politicians or state employees. However, there is no evidence that such relations exists, but they are common in West Africa, and are thus likely to be present in some of Mopti’s associations (Blundo 1995). Overall, it seems like civil society exists in Mopti, but it is not possible to depict how many associations it includes. Summarizing the entire assessment of the theory, all its theoretical arguments help explain why the cases do not fulfill all elements of substantial democratization. For instance, communal belonging and reciprocal and vertical relations contribute to illustrate people’s political participation and representation, because people instrumentalize these relations and belonging. In addition, the theory illuminates how actors obtain accountability and legitimacy in elections, and why political changes do not happen in the cases. The lack of changes is due to recycling of new and old political actors and non-separation of public and private spheres. Despite the theory’s good explanatory value on these arguments and others, the cases deviate from the theory’s arguments on some matters. For instance, the communal belonging is not important in voting in Mopti. The theory makes clear that this deviation, as well as other ones can be explained as resulting from a stronger presence of other of the theory’s traits in the cases. One example is the use of neo-patrimonial relations instead of communal belonging in Mopti’s politics. Despite this explanation, the theory does not clarify the underlying causes for why some traits are absent or do not play a significant role in the cases. Rudebeck (2001:87) illuminates this point in his criticism of the theory’s use of culture as the central, underlying explanatory argument for all the others discussed here. He argues that this use of culture tends to 99 be close to “tautological”-“by tending to explain political inefficiency (clientelism and corruption) with culture marked by clientelism and corruption”. By focusing on culture, the theory also downplays the importance of local power relations, leaving out key issues such as resource distribution and equality (ibid.:87). Even if distribution and equality are not intrinsic to democratization, they are vital components in democratization processes (ibid.). By leaving them out, the theory thus fails to capture crucial processes involved in democratization (ibid.). In all, Rudebeck’s critique as well as my critique above weakens the theory’s explanatory value regarding African states and their politics.

9.3 Local politics - a complex game The specific questions outlined from Bierschenk and Sardan’s theory (1998, 2001) that I will answer in this section are the following: -Is there more than one political power arena in the cases? -Do different arenas contribute to fragment political power in the cases? -Do development brokers exist? Do these brokers make municipal council’s power fragmented? Do brokers benefit from their role in terms of gaining resources? -Do village councils and municipal councils have a fluid character? Are decision- making oral and informal? -Do local governments have popular and institutional accountability? -Has the state gained control and legitimacy through creating a democratic space?

There are two power arenas in Mopti. The village chief, ward chiefs, and founding families constitute one arena and the local government another. Tambaga also has these two arenas, but the local government’s arena is weaker than the village chief’s one, because it has less legitimacy in the population than the village chiefs do. The village chiefs’ strong position demonstrates the difference between the two arenas in legitimacy (see sect 9.2). Compared to Mopti’s local government, Tambaga’s local government has less legitimacy in the population, because it only provides some services and does not engage in neo-patrimonial relations as in Mopti. In Kontela, village chiefs comprise the sole power arena, because Kontela’s local government 100 lacks legitimacy in the population. This is because it provides few services to the population, and large parts of the population do not know it exists (see sect 7.1.3). This argument of power arenas thus fits with Mopti and Tambaga, but not with Kontela, because here only one power arena exists. Despite this, the theoretical argument of several power arenas is fruitful, because in addition to explaining politics in Tambaga and Mopti, it illustrates the difference between the cases as to the result regarding substantial democratization. For instance, Mopti’s government performs better than Kontela’s, because in Mopti it constitutes a separate power arena. About the question of power fragmentation, this relates to the latter question of power arenas, because the existence of several power arenas implies fragmentation of power. Thus, there is fragmentation in Mopti and Tambaga, but also in Kontela, because there are several actors within Kontela’s sole power arena, such as local government and political parties. The latter two wield some influence over the village chiefs’ political power. There is however more fragmentation in Mopti than in the other two cases, because there are more actors involved in the two arenas in Mopti, due to presence of neo-patrimonial relations there. These relations involve the electorate. As to the theory’s argument about development brokers, they exist in Mopti, as small NGOs with educated persons who try to get funding from larger NGOs (NGO worker [interview]). As most other NGOs, they contribute to fragmentation of local government’s power. This is because they provide services to the population and thus obtain the population’s legitimacy instead of the local government. The theory is therefore descriptive of Mopti’s situation on this point. However, the role of these brokers and NGOs in fragmentation of local governments’ power is limited, since they are less important for the population than the clientelistic relations the population enjoys (see sect.9.2). Further, it cannot be documented whether the associations and NGOs profit from their position as brokers in Mopti, due to lack of information on such a sensitive issue. In Tambaga, there is one broker, who is a councilor for several NGOs, but he does not contribute to fragment local government’s power, since the NGOs he works for have few projects 101 in Tambaga. Nor does he benefit much from being a broker. In Kontela, there are no such brokers. The theory therefore describes the situation in Mopti and partly in Tambaga, but not in Kontela. As to the question of whether village councils and municipal councils have a fluid character, and their decision-making is oral and informal, this fits all the cases (see ch 7). This is because decision-making processes are characterized by consensus and oral discussions. In addition, in Mopti decisions are often agreed upon in informal meetings with several actors involved before municipal council meetings. Regarding local governments’ institutional and popular accountability, only Mopti’s and Tambaga’s local governments have achieved this to some extent (see sect 7.1.3). The theory thus partly explains these two latter cases’ outcome, and fully elucidates Kontela’s results as to substantial democratization. Further, about the question of the state gaining control and legitimacy through creating a democratic space, it has not happened in any of the cases. State legitimacy and control remains low, which fits the theory, arguing that the state is only one of many actors. Hence, on this point, the theory clarifies the cases’ result regarding substantial democratization. Concerning whether the state has created democratic spaces, it has indirectly facilitated establishment of one democratic space through the decentralization reform in all three cases. In Mopti, this democratic space is the council, because women have got representation on the council after decentralization. In the other two cases, the democratic space is in the political parties, because decentralization has lead to increased political participation in political parties. All in all, the theoretical arguments discussed in this section help explain the result of the assessment of the cases regarding democratization, as well as the variation between them. Nevertheless, even if the theory thus has good explanatory value, there are two failures to it. The first is that it is difficult to draw theoretical conclusions from it, because it is an empirical and contextual based approach. The second is that it does not provide underlying explanations for why its theoretical factors appear and endure, or are undermined over time. For instance, the theory argues that fragmentation is due to the 102 existence of more than one power arena. However, the theory does not outline how these power arenas have subsisted throughout time.

9.4 Comparison within cases In this section, I will undertake a comparison within cases over time to assess the theories’ ability to explain changes. To measure this ability, I compare the situation in the cases before decentralization to the one after, the latter described in the previous sections. I assess the theory on Tambaga and Kontela together, because they have many similar traits. The assessment on Mopti is separate, since it diverges from the other two cases on many points. I have outlined the theoretical questions used in these assessments below. Taken into consideration the context of the three cases before decentralization, only the most relevant factors for this period are included in these assessments. Based on Mamdani: -Were the cases in a state of centralized despotism? Did civil society have the same character as after decentralization? Were the cases together examples of a bifurcated state? Based on Chabal and Daloz: -Was there political instrumentalization of the communal belonging? Did the state instrumentalize village/ward chiefs? Were reciprocal and vertical relations instrumentalized? Did instrumentalization of goods and displaying wealth constitute the basis of political legitimacy and accountability? Based on Bierschenk and Sardan: -Was there more than one political power arena in the cases? Was political power fragmented? Did local institutions have a fluid character? Were decision-making processes oral and informal?

9.4.1 Kontela and Tambaga According to Mamdani, Mali was in a state of centralized despotism from independence and on, because it was among the states that abolished village chiefs’ institutions and replaced them with local party cadres of the single party (Mamdani 103

1996:25-26). These party cadres represented the state in rural areas, and they forced village chiefs to obey them. This worsened the rural-urban divide (ibid.). Concerning Tambaga and Kontela’s local governments, centralized despotism characterize these governments in the period before decentralization. These governments were examples of centralized despotism because the state was repressive, and enforced its rule through its local party cadres, primarily the state representative. However, the state also used village chiefs to govern (Tag 1994:102). It used the chiefs’ customary power status to gain legitimacy in the population, and forced the chiefs to carry out services, as tax collection (councilor [e-mail]; anonymous 17.10.03[interview]). Repression stopped in both cases with establishment of national democracy, and this reduced the state’s power locally. The end of repression also changed the position of the village chiefs, because the state could not force them to collect taxes. Further, end of repressive rule led to the populations’ refusal to pay taxes. As to civil society, it was tribalized before decentralization, because village chiefs influenced it, and its character has therefore not changed from before to after decentralization. All in all, although centralized despotism and a tribalized civil society existed in Kontela and Tambaga, I cannot answer whether a bifurcated state was present until having studied the urban case Mopti, because a bifurcated state implies that a rural-urban divide exist between local governments. I thus have to look at Mopti before stating whether such a state is present. Still, as to the points that I have discussed so far, the theory explains government and politics in the two cases before decentralization. Turning to the theory of Chabal and Daloz, instrumentalization of communal belonging was not present before decentralization, because there was little political activity in the two cases, according to informants. Also, the increase in political participation after decentralization was not entirely due to the communal belonging, but to an interest among people to partake in politics, as several informants expressed. People were interested because they thought that they could instrumentalize their participation; gaining contacts and resources by participating. Further, instrumentalization of reciprocal and vertical relations existed before 104 decentralization, since the few active politicians enjoyed neo-patrimonial relations with national level party cadres (Hanke 1999:99-100; Tag 1994:92). Still, despite these arguments fit the cases before decentralization, one type of instrumentalization by state representatives and politicians has disappeared after decentralization: instrumentalization of village chiefs. The local state representatives and politicians can no longer instrumentalize these chiefs. The theory cannot explain that instrumentalization of village chiefs does not exist anymore, because it argues that instrumentalization of these chiefs is a general feature of the disorder that is not likely to disappear (ibid.:162). Regarding the theory’s argument of distribution of goods and displaying wealth as the basis of political legitimacy and accountability, this does not apply to the cases before decentralization, only after. This is because the state enforced its rule through repression before decentralization. The theory does not provide the underlying causes for this change over time. Nor is it able to explain why the state’s instrumentalization of the village chief has been undermined. Hence, the theory only partly describes Kontela’s and Tambaga’s situation before decentralization, as well as the changes from before decentralization to after. Concerning the first and second questions on the theory by Bierschenk and Sardan, power arenas and power fragmentation, the two cases only had one power arena and no power fragmentation before decentralization. This was because power was concentrated around village chiefs and state representatives, who represented one power arena. Decentralization established another power arena in Tambaga: the municipal council. Hence, village chiefs got this arena to play on as well. In Kontela, the council was included in the existing power arena of the village chiefs, outlined in section 9.3. However, the council conditions village chiefs’ formal participation, i.e. its consultation with them, but not their informal, strong influence through the councilors. In addition, regarding whether local institutions had a fluid character and the decision-making processes were informal and oral, as the theory argues, village councils and the state were the only ones present and thus the ones to be discussed here. The state did not have a fluid nature, nor informal decision-making processes, locally, because it had a strong 105 centralized character. As to the village councils, they had a fluid character, and their decision-making processes were informal and oral, because they were informal structures where decisions were taken by consensus, as several informants stated. In all, on this point, the theory can only explain the character and decision-making processes of village councils. Moreover, the theoretical argument that the state lacked accountability and legitimacy is descriptive of the cases, because the state was neither accountable nor did it enjoy legitimacy in population, due to its repression and abuse of tax money (anonymous 17.10.03 [interview]; Tag 1993:39). In general, with the exception of not describing the local state’s character before decentralization, all the other theoretical factors of Bierschenk and Sardan help explain the situation before decentralization and illuminates changes over time. The most significant changes from before decentralization to after were the state’s reduced power, the village chiefs’ strengthened position, and the increase in political participation. The establishment of national democracy facilitated these changes; decentralization only reinforced them. However, as stated in 9.3, the theory of Bierschenk and Sardan does not outline the underlying causes for why these changes happen. The theory of Chabal and Daloz fails on this point as well. These weaknesses of the two theories reduce their explanatory value.

9.4.2 Mopti The differences in Mopti from before to after decentralization were of another character than in Tambaga and Kontela, because Mopti had been a municipality for a long time. Mamdani’s theory argues that in countries with centralized despotism, such as Mali, this system was present all over the country, but the rural-urban divide still existed. This implied that urban local governments, such as Mopti had civilized governments, used modern law, and the population was thus citizens. This applied to Mopti, but at the same time Mopti had a village chief, ward chiefs and founding families, who influenced local government (Massing 1996a). Further, customary law was utilized in disputes over natural resources, as well as in other conflicts (Bouju, 2000:151;NGO worker 13.11.03[interview]). It is likely that modern law was in use in some instances as well, but since there was authoritarian rule in Mali, equal 106 treatment before the law is not likely to have been ensured in Mopti. The simultaneous existence of local government and customary chiefs, as well as customary law and modern law makes Mopti in a contradictory situation before decentralization that the theory cannot explain. Still, since Mopti had a civilized local government and modern law, it differed from the other two cases, and there were aspects of a rural-urban divide between them. Consequently, elements of a bifurcated Malian state existed before decentralization. Introduction of national democratic rule in 1992 and decentralization in 1999 deepened this divide, as Mamdani argues that democratization and decentralization is likely to do (ibid.:298). This divide was aggravated because of changes in Mopti. One such change was that the power of Mopti’s village chief, ward chiefs, and founding families was weakened with establishment of new political parties from 1992. Their power was reduced, because they had to build up alliances with all the new parties. Under the single party rule, they enjoyed a stable alliance with the single party. With decentralization, the municipal council’s power was reinforced, because the council did not consult the village chief, nor the ward chiefs or the founding families in the decentralization process (Massing 1996a). Overall, even if Mopti was in a state of centralized despotism before decentralization, and elements of a bifurcated state existed between the three cases, the theory does not clarify Mopti’s contradictory situation, having a village chief and ward chiefs, as well as customary law. This is because the theory argues that neither did these chiefs and this law exist before decentralization, nor after in urban areas such as Mopti. Turning to Chabal and Daloz, their argument about instrumentalization of the village chief and ward chiefs does not apply to Mopti, since these chiefs were not instrumentalized by political actors, but actively participated in Mopti’s politics through the neo-patrimonial relations they enjoyed with the municipal council (Bouju 2000:147). The situation thus differed from the two rural cases, where instrumentalization of such chiefs existed. Further, both chiefs and councilors in Mopti instrumentalized the neo-patrimonial relations in politics to gain resources, such as property (Bouju 2000; Massing 1996a). Introduction of multiparty system 107 and decentralization made these relations more complex, because several political parties got involved. In addition, multiparty rule and decentralization signified more autonomy for the municipal council, as previously discussed in this section. The council took some decisions without consulting the village chief and ward chiefs. Although this implied that the village chief, ward chiefs, and founding families had fewer maneuvers for making instrumental gains, they were still significant actors after decentralization because they had sanction powers over council decisions in many instances (see sect 7.1.2). Regarding the theoretical aspect of the communal belonging, it determined political participation and representation together with reciprocal and vertical neo-patrimonial relations before decentralization. The communal belonging included the status as being of the founder families of Mopti. People’s belonging to the founder families or not was significant, because people from these families were the ones to have the village chief and ward chiefs positions (Bouju 2000). People in these positions thus instrumentalized it in politics, together with the neo-patrimonial relations they enjoyed, because members of the founding families were patrons in the neo-patrimonial relations. Hence, the factors of instrumentalizing one’s belonging to the founding family, i.e. one’s communal belonging, as well as instrumentalizing reciprocal and vertical neo-patrimonial relations were present in Mopti, and these traits thus explains Mopti’s politics. The argument about distribution of goods and displaying wealth as constituting the basis of political legitimacy and accountability explains Mopti’s politics before decentralization, because material goods, such as property, were important elements in the political instrumentalization of neo-patrimonial relations. These relations constituted the foundation of political legitimacy and accountability for all actors in Mopti. Despite the council’s attempt to undermine these relations, they are still important for accountability and legitimacy, since the village chiefs and founding families involved in the relations have sanction powers over the council’s decisions. Hence, this theoretical factor, which describes the basis for accountability and legitimacy, explains Mopti’s politics before decentralization. Pertaining to the argument of fragmentation of political power that 108

Bierschenk and Sardan propose, such fragmentation was present in Mopti’s politics before decentralization, because several political power arenas existed (Bouju 2000). These arenas were the state administration, the municipal council, the village chief, the ward chiefs, the founding families, and the chiefs of natural resources. Decentralization did not make any of them disappear or increased the number of such arenas. However, with the exception of the local government arena’s influence, which was strengthened the political influence of the other arenas was reduced. Regarding the arguments that local institutions have a fluid character and decision- making processes are informal and oral, it is relevant to discuss the local government, since this was the key institution in Mopti. It had a fluid character, because of the influence of neo-patrimonial, informal relations, which made decision-making processes oral, just as the theory argues. In sum, Bierschenk’s and Sardan’s theory helps explain Mopti’s situation before decentralization and the changes that have appeared with decentralization. Mamdani’s theory, though, describes the situation before decentralization, because elements of a bifurcated state existed. However, it cannot clarify Mopti’s contradictory situation. The theory of Chabal and Daloz elucidates Mopti’s politics before decentralization, with the exception of instrumentalization of village chiefs and ward chiefs that were not present before decentralization. This theory also explains changes over time, except for the changes that have appeared in Mopti’s council regarding the weakening of instrumentalization of neo-patrimonial relations.

9.5 Generalization of findings Despite variation between the cases, it is fruitful to explore whether some of their traits are general for all Mali’s local governments, or whether the two rural cases are representative for all rural governments, and Mopti for all urban ones. This section compares findings from the assessment of the theoretical approaches in the previous sections (except 9.4) to all of Mali. Starting with a discussion of Mamdani’s arguments regarding the urban-rural divide and the bifurcated state, customary chiefs exist and are important political actors all over Mali, both in urban and rural 109 municipalities (Beridogo 1997; Coll 1997;Njienhuis 2003; Kassibo 1997). In some places, decentralization has strengthened the village chief’s political role (Béridogo 1997; Nijenhuis 2003), while in other places the reform has weakened its power (Massing 1996b). This variation in the village chief’s power means that some places they may resemble decentralized despots, while other places not. Because I cannot rule out that decentralized despots may exist in Mali, Mamdani’s theory is partly right on this point. Even if many of Mali’s village/ward chiefs are not such despots, they are nevertheless important political actors in most Malian municipalities. However, because village/ward chiefs are also present in urban municipalities, the theory partly fails, because it states that such chiefs are only to be in rural areas. It cannot explain why they exist in urban ones as well. As to usage of customary law, it is not possible to depict a general picture for Mali, due to lack of generalizing studies. However, some case studies show that customary law is in use in other cases (Ba 2002; Le Marcis 1999; Vedeld 2002). Further, it is likely that customary law is applied in the entire country, because the local government act recognizes its importance in solving conflicts at the municipal level. However, neither is it possible to say anything about the usage of this law in urban local governments, nor about the application of modern law in urban areas. Consequently, I cannot generalize on whether Mali has elements of a bifurcated state or not. As to Mamdani’s argument regarding the lack of balance between decentralization and centralization, autonomy and alliance, and participation and representation, which is also to explain why decentralization does not result in democratization, this argument is valid for the entire country. The lack of balance between these aspects has the same reasons as in the three cases assessed in the thesis. The cases are thus representative on this argument. Regarding balance between decentralization and centralization, it is not fulfilled, because the slow national devolution process makes all municipalities in a state of centralization rather than decentralization. Moreover, as to autonomy and alliance, the three cases are representative, because there is no balance between these aspects. This is because of the refusal to pay taxes, which is present in the entire Mali (UNDP 2003:26). 110

All municipalities thus rely on the state. The refusal is however more widespread in urban municipalities than in rural ones (ibid.). As to balance between participation and representation, no statistics on political participation exists, and it is therefore not possible to say whether there is a balance or not. However, regarding certain patterns, the cases present a general picture of Mali. For instance, the weak political representation of women in the cases applies to the country, because only 16% of Mali’s councilors are women (UNDP 2003). As to the caste system’s political importance, it varies between municipalities, as the three cases studied in this thesis demonstrate. Summarizing Mamdani’s argument on the lack of balance, it helps explaining Mali’s overall situation concerning democratization. Turning to the theory of Chabal and Daloz, the latter paragraph shows that the caste system, which is included in people’s communal belonging, is instrumentalized in political participation and representation in Mali. However, the degree to which it is instrumentalized may vary between municipalities. In elections, for instance, the caste system is a vital component in several Malian cases (Béridogo, 1998, Kassibo 1997; de Langen 2003). Also, instrumentalization of village belonging, which is another element the communal belonging embraces, is stronger than party loyalty in elections (Le Marcis 1999:162). However, according to several informants, in some municipalities, many castes participate in political parties and are represented on municipal councils. Their mobilization reduces instrumentalization of the caste system/caste belonging in politics (Fay 2000:136), and thus the communal belonging as well, since caste is part of a person’s communal belonging. The theory’s explanation to this latter reduction is that the other factors of the theory are more important than the communal belonging. This is true, because the factor instrumentalization of reciprocal and vertical relations exist all over Mali (Beridogo 2002:30-31; Bouju 2000). As to recycling of political elites, several other cases in various regions depict recycling (Tag 1994:92; Béridogo 1998; de Langen 2003). Concerning the argument of actors’ instrumentalization of civil society, it is not possible to find a general picture, because of no information about it. Moreover, as to non-separation between public and private spheres, this is true for the entire 111 country (Lange 1999), due to rent-seeking practices in central state administration. Even if the theory thus fits with Mali on all points, it has two failures. The first is that people do not instrumentalize village/ward chiefs, because these chiefs have a persistent role most places. The second failure is that it does not outline the underlying causes for why political instrumentalization of the communal belonging has been lessened. Altogether, even if the theory fails to illuminate the causes for these two deviations, the other factors of the theory are present in Mali’s local politics. Hence, the theory helps explain politics in Malian local governments. As to the theory of Bierschenk and Sardan, their argument on fragmentation in local politics is descriptive of Malian local politics in general, because in most municipalities the two local power arenas of local governments and village/ward chiefs are likely to exist. The precise number of power arenas varies, and in some cases, there might be only one, such as was the case in Kontela. Further, informal, oral processes characterize decision-making in other Malian cases as well (Beridogo 1998; de Langen 2003), but I cannot document whether such processes are general for all of Mali. As to whether local governments have achieved institutional and popular accountability, it is not likely, because of the general problem of scarce tax revenue and councilors’ lack of information on decentralization (UNDP 2003). Concerning civil society, I cannot say whether development brokers are present other places, because no information exist on this issue. In sum, the arguments of Bierschenk and Sardan discussed here fit to the entire country. However, the theory’s two weaknesses regarding its explanatory value outlined in 9.3 apply to the entire country as well. Overall, after having discussed the three theories in this section, it is not possible to conclude that democratization has failed to materialize in the entire country, because not all factors of the theories were present or possible to document. However, at least some challenges to democratization, such as the presence of village chiefs, accounts for the entire country. 112

9.6 Conclusion This chapter has assessed three theories’ ability to explain why only some elements of substantial democratization have materialized in the three cases. Together, the theories provide many relevant explanations. As to Mamdani’s theory, it gives some explanations to the cases’ outcome, since elements of a bifurcated state exist in the cases before and after decentralization. However, the theory cannot explain Mopti’s contradictory situation before and after decentralization, nor the absence of decentralized despotism in the rural cases after decentralization. With the exception of the strong presence of village chiefs, it is not possible to generalize on these findings, due to no literature on the theory’s aspects. As to Mamdani’s second argument, the lacking balance between decentralization and centralization, participation and representation, autonomy and alliance, it fits the cases, as well as the entire country, and thus contains relevant explanations for why local governments are not accountable and representative. The factors of Chabal’s and Daloz’s theory that elucidate the cases’ outcome are instrumentalization of communal belonging, reciprocal and vertical relations, as well as non-separation of public and private spheres and recycling of political actors. Still, the theory fails to outline the underlying arguments for why the significance of the communal belonging is reduced over time, as well as why village chiefs are not instrumentalized in politics in the rural cases after decentralization, only before. The theoretical traits that I have information about for the entire country are however descriptive of local politics and government in Mali. As to the theory of Bierschenk and Sardan, most of its factors explain the cases’ outcome, because fragmentation of council’s power, existence of several power arenas, informal decision-making processes, and lack of state control and legitimacy exist. However, the theory does not give details on the lack of fragmentation in Kontela and Tambaga before decentralization. Additionally, it has two failures: the first is the lack of general, theoretical arguments one can draw from it. Its other failure, which it shares with the theory of Chabal and Daloz is that it does not explain why some of its factors are not present or are reduced in significance over 113 time. In general, because none of the theories gives comprehensive explanations for the result of the cases regarding substantial democratization, the theoretical challenge is to develop contextual, theoretical factors that can explain why not all elements of substantial democratization have been fulfilled in the three cases. 114

10. Concluding remarks This thesis has demonstrated that Mali’s democratic decentralization reform has contributed to realize some elements of substantial democratization in the three local governments studied. It has also illuminated the varying impact such a reform has had on the three local governments. Hence, the thesis has provided new understandings of the effect of a democratic decentralization reform in diverse contexts, as well as the importance of local social and political processes for the result. This chapter sums up the main conclusions of the thesis. It also briefly presents democratic decentralization experiences in other countries, which further illuminates why full democratization has not materialized in the three Malian cases. At last, the chapter elaborates on future research issues.

10.1 Main findings, explanations and conclusions This thesis has studied Mali’s democratic decentralization reform, concentrating on democratization of municipal level government, because at this level, people encounter the new decentralized institutions. In order to study Mali’s democratic decentralization reform, the thesis has assessed three cases in terms of whether substantial democratization has materialized in them, and if it has, to what extent. The assessment has used Törnquist’s substantial democratization definition and framework, as well as the variables of accountability from the framework of Crook and Manor and aspects from Blair’s assessment of democratic decentralization. Out of many African democratic decentralization reforms, I have focused on Mali’s reform, because it represents a critical case of democratic decentralization for several reasons outlined in chapter three. Besides, the three local governments discussed were critical cases and the choice of them was based on the three theories I have used. The two research designs I have used to study the three cases; most similar and most different research design, have been fruitful in this thesis. This is because they have facilitated a systematical analysis of the similarities and differences between the cases after decentralization, and within them from before decentralization to after (see ch 7,8,9). The designs have also been good tools for 115 testing the theories’ explanatory value on the cases and within them. For instance, the most different design has helped elucidating that Mopti’s local government was more accountable than the other two. Moreover, the most similar design has contributed to shed light on changes that have taken place within the cases over time. For example, this design has helped depicting that the local governments of Tambaga and Kontela changed from being in a state of decentralized despotism before decentralization, to being replaced by elected local governments after decentralization. Here is the main result of the assessment of substantial democratization: Figure 10.1: Assessment of substantial democratization

MOPTI TAMBAGA KONTELA

Low quality and Low quality and narrow STATE Mediocre quality, wide scope narrow scope scope

Mediocre quality, Low quality, narrow GOVERNMENT Mediocre quality, wide scope narrow scope scope

Judiciary partly exists, and Non-existing JUDICIARY, has a mediocre quality. Non-existing judiciary, no judiciary, no CITIZENSHIP Judiciary and citizenship has citizenship citizenship good scope

All respected, except All respected, except POLITICAL AND All respected, except equal equal treatment equal treatment before law CIVIL RIGHTS treatment before law before law & freedom of assembly

Low quality, wide CIVIL SOCIETY Mediocre quality, wide scope Low quality, wide scope scope

As summarized in this model, all three cases share the same outcome: substantial democratization has not fully materialized in any of the cases. Despite this, the model shows that the cases vary regarding the extent to which their institutions and rights have achieved high quality and wide scope. Mopti’s state administration and its local government have mediocre quality, because it is not representative, nor open, and only partly accountable. The influence from neo-patrimonial relations on the local government’s decision-making, and the clientelistic practices state employees engage in explains this result. However, the scope of the state administration and local government is wide, because they reach out to the entire population and provide many services. Mopti’s civil society is somewhat democratically oriented, because 116 only some of its associations have an impact on local government. As to rule of law, judiciary, and citizenship, they all have low quality in Mopti, since Mopti’s judiciary applies both customary law and modern law. Mopti’s judiciary has a wide scope, though, because people can easily get in touch with it. Use of customary law implies that the civil right of equal treatment by the law is violated. With the exception of this right, all other civil and political rights are respected. Regarding the cases of Tambaga and Kontela, the model shows that their state administrations have low quality and scope, because they hardly exist in the two cases, since there is only one state representative in each place and they are seldom in contact with the population. Moreover, Tambaga’s local government has mediocre quality, because it is partly accountable to the electorate, but not representative nor open. As in Kontela, the local government’s scope is narrow, because people are not in frequent contact with it, and it only partly reaches out to people with services. Kontela’s local government has low quality, because it is neither open, nor accountable or representative. Further, judiciary and rule of law is not present in Kontela and Tambaga, because customary law is used. The civil right of equal treatment before the law is therefore violated, and citizenship does not exist. All other civil and political rights than equal treatment are respected, except for the right to assembly in Tambaga. Also, Kontela’s and Tambaga’s civil society has a wide scope, since most people are members of associations. Still, it has low quality, because only some associations influence local government. It is therefore not democratically oriented in any of the two cases. Moreover, decentralization has led to increasing political participation in Kontela and Tambaga. However, only in Kontela, it has resulted in representation on council for one other caste, in addition to the already existing political elite (nobles). Consequently, there is no equal distribution of seats among caste groups in these two cases. Actors’ capacity also differs between the cases. The ones identified as political actors were state representatives (administration in Mopti), civil society, village/ward chiefs, municipal councilors, and members of political parties. I also assessed the population’s capacity as an entity. The assessment showed that in all three cases, there are variations between 117 actors’ capacity. Generally, the village chiefs are the ones with the highest capacity, because they link activities in three parts of the political terrain, and politicize by single issues and communal loyalties. They are also capable of mobilizing the population, as well as political parties and councilors. Furthermore, the findings demonstrate that the actors who politicize on single interests and mobilize by incorporation through neo-patrimonial or clientelistic relations are the ones with the highest capacity. For instance, in Mopti neo-patrimonial relations, and thus incorporation, are crucial in determining actors’ capacity. If the village chief, the founding families, the political parties and the councilors do not use these, they cannot mobilize the electorate. Political parties and councilors stand few chances to gain representation on the municipal council if they do not use these neo- patrimonial relations. Consequently, politics evolve around single issues and actors relate to institutions and rights by promoting their own instrumental aims. They thus use institutions and rights, and sometimes abuse them as well, exemplified by the sales of property in Mopti. In Tambaga and Kontela, political parties and councilors do not have as much capacity as the ones in Mopti, because the village chiefs distribute property instead of the local governments. This makes local government less important to collaborate with for the village chiefs. Moreover, even if councilors and political parties mobilize people by neo-patrimonial relations in Kontela and Tambaga as well, these relations only endure during elections. As to the population in general, it thus lacks the capacity to mobilize outside neo-patrimonial relations and use institutions and rights in all three cases. This constitutes a severe problem, and illustrates why the cases have not yet achieved full substantial democratization. Still, even if the strategies actors employ discussed in this section explain the varying capacity of political actors, the frameworks present additional conditioning factors that determine actors’ capacity. These factors are the political opportunity structure, habitus and the three forms of capital actors possess, as well as their ability to transform the capital forms into symbolic capital/political legitimacy. The aspect of opportunity structure explains the variation in actors’ capacity within and between the three cases. For 118 instance, women and youth have restricted political opportunities in Kontela and Tambaga, compared to middle-aged noble men, while in Mopti women have the opportunity to participate in politics. Further, habitus, which may be defined as social norms, illustrates people’s varying capacity to be elected to Kontela’s and Tambaga’s councils, because only nobles are considered eligible. In Mopti, neo-patrimonial relations between electorate, customary powers and political parties determine representation and not habitus/social norms. Hence, habitus only contributes to illuminate actors’ capacity in Tambaga and Kontela. The third factor- possessing different forms of capital and being able to transform these into symbolic capital- comprises the most important explanatory factor for actors’ varying capacity in all three cases. For instance, Mopti’s village chief has social and cultural capital that he transforms into symbolic or political power. Likewise, Mopti’s councilors have economic capital that they convert into political power, through neo- patrimonial relations with village chiefs, founding families and electorate. In Kontela and Tambaga, village chiefs and councilors are the only actors who are able to transform this capital into symbolic capital, but the councilors do not have economic capital such as in Mopti. This explains the greater importance and capacity of councilors in Mopti compared to the other two cases. Altogether, the factors of opportunity structure and capital and its transformation into political legitimacy provide explanations of actors’ varying capacity in all three cases, while habitus only illuminates actors’ capacity in Kontela and Tambaga. What are the further explanations for the cases’ outcome? In this thesis, I used both general explanations based on the two analytical frameworks and contextual and theoretical explanations based on three theories. First, I will repeat the general explanations by the frameworks of Törnquist and Crook and Manor, discussed in chapter eight. These explanations partly overlap with each other, but Törnquist’s explanations specify the context factor, which Crook and Manor outline. The latter authors focus on the importance of the local social and political context for democratization. This applies to the cases, and explains the elite capture of local government in Tambaga and Kontela. Further, the context illuminates the differences 119 between the cases in performance of institutions. For instance, Mopti’s local government has higher quality than the governments of the other cases, because of its long experience of local government and strong state presence. The social context also matters in the cases, which is illustrated by the caste system’s dominant role in political representation on the councils of Kontela and Tambaga. In addition, village chiefs’ strong positions in all three cases and Mopti’s neo-patrimonial system demonstrate the significance of the social context for the result of the cases. All these factors help explain why none of the local governments is open or representative. They also show why Tambaga’s and Mopti’s local governments are partly accountable and Kontela’s government not accountable at all. The third factor that explains the cases’ outcome is available resources. There are few resources in all cases, which elucidate why none of the three local governments has high accountability. The fourth factor Crook and Manor outline is presence of competitive political parties. There are no competitive parties in the cases, which provide an explanation for lack of democratization, because there are no parties to enforce the accountability of the three local governments. A fifth factor is the widely distributed free press, which only exists in Mopti, and demonstrates the difference in accountability between the three local governments. The sixth factor is professional civil service, which is not present in any of the cases, and provides an additional explanation for the outcome found. A seventh factor is a vibrant civil society. All three cases have such a civil society, although its character varies between the cases. Only in Mopti does the civil society have the capacity to promote democratization. Turning to Törnquist’s explanations, they specify the factor of context outlined by Crook and Manor, because all the four explanations relate to the local political and social context. The first is that elite-based reform rarely ends in popular democratization, because such a reform does not involve the population. This fits the cases, because the reform only engaged the state representative and village chiefs and other people in the political elite in Tambaga and Kontela. In Mopti, the reform involved the local government and the state administration. The second factor Törnquist outlines is 120 incapacity to link activities in different parts of the political terrain. The population of the three cases was not capable of using institutions and rights, and this factor thus contributes to explain why not all elements of substantial democratization have been fulfilled. The third factor is politicization of single issues, resulting in domination of clientelism in politics. Single issues and clientelism applies to the three cases, previously discussed in this section. The fourth factor is non-existence of public spaces where one can mobilize political issues, which gives an explanation for lack of representative and open government in the cases. This is because even if public spaces exist in the cases, they only contribute to political mobilization in Mopti. In the other two cases, social norms exclude women and youth from participating. Altogether, Törnquist’s general explanations are valid for the cases’ outcome. Even if all these explanations are valid, they are general ones and can be used on all decentralization reforms throughout the world. There is thus a need for contextualized explanations that deal specifically with the African context. The thesis has assessed the explanatory value of specific theoretical arguments derived from three theories. These theories have different explanations for why decentralization does not result in democratization in Africa. As to one of the theories used, that of Chabal and Daloz, it explains lack of democratization from decentralization as owing to two interrelated factors: instrumentalization of communal belonging and reciprocal and vertical relations. These lead again to other traits, such as non-separation between public and private spheres, instrumentalization of traditional sphere. According to the theory, decentralization is not likely to lead to any changes in these traits. The assessment of the theory in chapter nine shows that the theory provides some explanations for the cases’ outcome. For instance, instrumentalization of reciprocal and vertical relations is present in the cases and impinges on the democratization process, because people vote according to these relations. Such relations also explain the pattern of representation on the council in the three cases. In addition, this aspect illustrate why Mopti’s local government and state is not open, nor representative, and only partly accountable. In Tambaga and Kontela, the communal belonging as in belonging to a caste better elucidates the lack of open, 121 representative and accountable government, as well as the voting pattern. This is because political representation on the council is largely reserved to the dominant political and social elite: the nobles. In Mopti, reciprocal relations have to be used instead. Moreover, as to the cases’ situation before decentralization, the theory fits on all points, with the exception of two instances: absence of instrumentalization of village and ward chiefs in Mopti and non-importance of communal belonging in politics in Kontela and Tambaga before decentralization. Regarding the assessment of the theory on the entire country, it applies on some factors, such as to some extent instrumentalization of the caste system; that is the communal belonging. Further, the theory accounts for variations between countries or cases concerning the importance of its factors (ibid.16). Despite all its strengths, the theory has two failures. The first is that it does not present the underlying causes for why such variations exist. Rudebeck (2001) illuminates this in his criticism of the theory as “tautological” (ibid: 87) in its explanations. The theory’s second failure is that it cannot explain why some factors are not important in the cases anymore. It argues that all its factors are general for Africa (ibid.:xix), and likely to persist for a long time. This is because “there is an interlocking neo-patrimonial logic” between politicians and the electorate (Chabal and Daloz, 1999:162). Owing to this argument, it cannot explain why communal belonging was not present in Kontela and Tambaga before decentralization. Nor does the theory give any account for why reciprocal relations have replaced the factor of communal belonging in political participation in Mopti. Also, the theory does not explain why there is no instrumentalization of village/ward chiefs in any of the cases after decentralization, only before. In addition to the critique posed by Rudebeck (see ch 9), the two failures listed here weaken the theory’s explanatory value. Concerning Mamdani’s theory, it has two main arguments for why decentralization does not result in democratization in Africa; the bifurcated state and the lack of balance between decentralization and centralization, autonomy and alliance, and participation and representation (see ch 8 and 9). The first, the bifurcated African state, implies that there is a rural-urban divide 122 between local governments, where people are citizens, governed by modern law in urban areas. Rural local governments constitute of decentralized customary despots with monopoly of power, who rule by customary law, thus making people subjects. The assessment of this theory on the cases after decentralization demonstrates that there are only some elements of the bifurcated state in the three cases, because decentralized despots do not exist in the rural cases. The elements of a bifurcated state present are customary law and strong village chiefs. A reason for why only some elements exist is that customary law, customary chiefs, and a tribalized civil society are present in the urban case Mopti. Mamdani does not explain Mopti’s contradictory situation after decentralization. Altogether then, this argument of the theory is only partly able to explain the cases’ outcome. It explains why Mopti is better off than the other two cases as to accountability, since it has elements of civilized rule, due to use of modern law, having an elected local government and a partly civilized civil society. Still, since only some elements of a bifurcated state exist, the theory does not provide a comprehensive explanation for why the cases differ on the various elements of substantial democratization. Regarding the situation before decentralization, the cases were in a state of centralized despotism, because local single party cadres wielded power in the local governments. The theory is thus descriptive of the situation before the reform. As to the theory’s second argument of the lacking balance between the previously mentioned aspects, this makes clear some of the causes for the cases’ result regarding substantial democratization. This is because a balance between decentralization and centralization, autonomy and alliance, and participation and representation does not exist. As to decentralization and centralization, there is no balance, because devolution and deconcentration has not been completed. Further, since they are all dependent on state transfers, this leads to municipalities’ alliance rather than autonomy. There is also lack of balance between participation and representation, because all castes, ages and both gender participate in political parties, but voters elect only middle age men and nobles to Tambaga’s and Kontela’s councils (except 2 ex-slaves in Kontela). In Mopti, neo-patrimonial relations contribute to lack of balance between participation 123 and representation. Altogether, the theory has some explanatory value for the cases’ outcome, because as long as elements of a bifurcated state are present, decentralization will not lead to democratization. Concerning the third theory, the one by Bierschenk and Sardan, all its factors apply to the three cases after decentralization, and the theory thus helps explain why not all elements of substantial democratization have materialized in Mali. Its arguments on the existence of several power arenas and hence fragmentation of power counts for the cases. Because several power arenas and fragmentation of power exist, local governments and state administrations have not obtained legitimacy and accountability in the cases. These factors further explain that the village chiefs are still strong in the cases, because they have maintained their separate power arena. Assessed on the situation before decentralization, the theory fits with Mopti’s local politics and institutions, because here several power arenas were present. However, it does not explain Tambaga’s and Kontela’s circumstances, because these cases only had one power arena, and no fragmentation of power before decentralization. As to the assessment of the theory on the entire country, it fits on most points. Altogether, the theory’s strength is that it explains why there is variation between and within the cases. The other two theories concentrate on general features for all African states, and only explain variations by stating that they allow for differences between cases, despite being general. However, the theory by Chabal and Daloz and the one by Bierschenk and Sardan share one weakness: they do not give an account of the underlying causes for why the factors of their theory are present or not. Furthermore, the emphasis on variation between cases by Bierschenk and Sardan also makes it difficult to outline theoretical arguments from it, because the theory is based on contextual and empirical arguments. On the other side, the theory of Chabal and Daloz is so general that most scholars can agree on some of its elements, such as the existence of reciprocal relations (see sect 8.2.1). Still, scholars differ on how to explain the presence of these factors (cf. Rudebeck’s comments, ch. 9). All in all, despite the theories provide many relevant explanations for the cases’ outcome, some of their explanatory factors fail. There is thus a need for 124 contextualized theoretical factors yet to be developed. Summarizing this section, there are many factors that explain the cases’ result regarding substantial democratization. Some apply to the entire country as well. Elite capture of local government, neo-patrimonial relations in politics, the caste system’s importance for representation in local government, village chiefs’ strong role and sanctioning power, lack of widely distributed press and general information on the reform in the rural cases are all aspects explaining the cases’ outcome. Further, other explanations which also account for the entire country are deficiency in devolution of services, tax revenue, infrastructure, competence among councilors and employees, state representation, state employees’ resistance towards decentralization. In order for decentralization to succeed, it is vital that local governments and the state deal with all these factors listed. In particular, equal participation and representation of all social groups, as well as devolution and deconcentration are crucial conditions for substantial democratization to fully materialize. The next section will demonstrate whether these factors also constitute challenges in other decentralization reforms around the world.

10.2 Experiences from other countries After having outlined the findings and conclusions of the thesis, I will turn to other experiences in the world that further elucidate Mali’s outcome. The comparison concentrates on two aspects: Mali’s decentralization compared to other countries and comparison between Mali’s shortcomings and other countries’ experiences. The second includes examples of necessary conditions for democratization to follow from decentralization. Considering the West African countries Senegal and the Ivory Coast, they scored substantially higher than Mali on an assessment of all African decentralization reforms (Ndegwa and Grandvoinnet 2002). One explanation is obviously that they have longer experience of local government. Senegal embarked upon democratic decentralization in 1972, but the municipal councils have been weak compared to religious leaders and the reform has not yet materialized in democratization (Beck 2001:603). In the Ivory Coast, participation has increased from decentralization, but little responsiveness from local 125 government subsists, and “public accountability was virtually non-existent” (Crook & Sverrison 1999:51; Olowu 2001). As to South Africa, it has one of the highest scores on the decentralization assessment mentioned above, but it is marked by“complete disengagement between local government and community structures” (Heller 2001:144). On an overall basis, lack of resources and poor infrastructure, as well as difficult living conditions hinder participation in Africa, especially by women and poor (Olowu and Wunsch 2003). Experiences on other continents reveal various results and challenges of decentralization. In Thailand and the Philippines, decentralization has resulted in political-economic gains by local bosses (Mc Vey 2000; Sidel 1999). Turning to the examples of necessary conditions for decentralization to result in democratization, the three most reputed and successful cases of democratic decentralization in the world help explain the cases’ outcome. These are the Indian state Kerala, the Brazilian municipality Porto Allegre and the Brazilian state Ceara. One explanation to the result in Kerala and other Indian states, as well as in Porto Allegre, accountability was achieved by decentralizing responsibility over budget allocation and other matters to “direct democratic forums” (Abers, 2000; Törnquist, 2002;Mathew, 2003:54-55). In these forums, people deliberated and decided on public issues. These actions contributed to widespread popular participation and thus democratization (Harriss et al, 2004). However, political parties with mass organizations were crucial in their subsistence, initiating and supporting them (Cameron 2003; Mathew 2003). A third explanation, which accounts for Kerala and Porto Allegre was that they both had a strong civil society capable of mobilizing people and spreading information about the reform (Abers 2000; Tharakan 2002). In Ceara, the state employees informed the population. Altogether, these factors illuminate important aspects that contribute to explain why substantial democratization has not fully materialized in the three cases studied in Mali. 126

10.3 Future issues of research Since few studies on decentralization’s impact on democratization at Mali’s municipal level exist, there is an obvious need for generalizing studies as well as in- depth case studies of this issue. Among all the challenges and deficiencies of the reform listed in this thesis, there are three aspects that have had a crucial effect on the three cases’ outcome and on the entire Mali. These are the ones on which future research has to concentrate. These aspects are the interaction between local government and village and ward chiefs, the influence of these chiefs as well as the caste system on people’s political participation in elections and their impact on people’s representation on local government.

10.4 Final conclusion This thesis has shown that the democratic decentralization reform in Mali has contributed to establish the formal institution of local government, as well as having fulfilled some civil and political rights. Even if these aspects are common for the cases, the quality and scope of all institutions and rights assessed in this thesis varies between the cases. Furthermore, substantial democratization requires far more than the mere existence of institutions and rights. There are thus still severe deficits to substantial democratization in all three cases, such as that the three local governments are not open, nor representative, and only partly accountable in Mopti and Tambaga, and not accountable at all in Kontela. Further, people do not have the capacity to use institutions and rights in any of the three cases. There are many causes and explanations to these deficits, but the crucial ones are elite capture of local government, the caste system’s role in determining political representation on municipal councils, and the existence of neo-patrimonial relations. Mali’s decentralization reform has not dealt with these problems. The most important factor in order for substantial democratization to materialize from Mali’s reform is that local power relations have to be restructured. This requires a strong local government and state capable of mobilizing people to take part and be represented in local government and other local forums. 127

References

Abers, Rebecca (2000): Inventing Local Democracy. Grassroots Politics in Brazil. Boulder: Lynne Rienner Publishers.

Adcock, Robert and David Collier (2001): “Measurement Validity: A Shared Standard for Qualitative and Quantitative Research” in American Political Science Review 95 (3) :529-545.

Ake, Claude (2000). The Feasability of Democracy in Africa. Dakar : CODESRIA.

Amundsen, Inge (2001): “The limits of clientelism”, Forum for Development Studies 28 (1):69-75. Oslo: NUPI.

Andersen, Svein S. (1997):Casestudier og generalisering. Forskningsstrategi og design. Oslo: Fagbokforlaget.

Anonymous (2004.08.09) Hilsen fra Kontela [e-mail to Siri Bjerkreim Hellevik] - accessible from [email protected]

Ba, Boubacar (2002):”Natural Resources and Decentralisation: Local Institutions’ role in co- management and local conflicts resolution in the Mopti region”. Paper presented on the I.A.S.C.P conference, Bloomington, May31-June 4, [online] (reading date : 10.04.04) URL : http://dlc.dlib.indiana.edu/documents/dir0/00/00/05/64/dlc-00000564-00/boubacarb/050300.pdf Conference paper.

Badie, Bértrand (2000): The Imported State- the Westernization of the Political Order. Stanford : Stanford University Press.

Bako-Afari, Nassirou (1998): « Relations de processus de décentralisation et pouvoirs traditionnels : typologies des politiques rencontrées ». (05.04.04) [online]URL : http://www.fao.org /sd/frdirect/Roan0014.htm

Bayart (1996): L’Etat en Afrique. La politique du ventre. Paris: Fayard.

Beck, Linda (2001):”Reining in the Marabouts? Democratization and local governance in Senegal”, African Affairs 100:601-621. Oxford: Oxford University Press.

Béridogo, Bréhima (1997):”Processus de Décentralisation au Mali et Couches Sociales Marginalisées”. APAD Bulletin 14: 21-35. Hamburg: Lit Verlag.

(1998):”Compétition des acteurs sociaux pour le contrôle du pouvoir et des resources dans la commune rurale de Zégoua (Mali). APAD Bulletin 16:99-112. Hamburg.Lit Verlag.

(2002):”Le regime des castes et leur dynamique au Mali” in Recherches Africaines, Annales de la Faculté des Lettres, Langues, Arts et Sciences Humaines, Université du Mali, Revue Semestrielle 00. Bamako:Université du Mali.

Bértrand, Monique (1992): « Un an de transition politique, de la revolte à la III République malienne”, Politique Africaine 47:9-22. Paris : Karthala.

Bértrand Monique(1999a): « Les élections communales maliennes de 1998, première edition:étirement electoral et remue-ménage partisan ». Politique Africaine 72: 212-220. Paris: Karthala. 128

Bierschenk, Thomas & Jean-Pierre Olivier Sardan (1997): “Local Powers and a Distant State in Rural Central African Republic”. Journal of Modern African Studies 35 :441-468.Cambridge: Cambridge University Press.

(eds.)(1998):”Les arenas locales face à la décentralisation et à la démocratisation,” chapter 1 in Bierschenk, Thomas and Jean Pierre Olivier Sardan (eds.):Les pouvoirs au village. Le Bénin rural entre démocratisation et decentralisation. Paris: Karthala.

Bierschenk, Thomas, Jean-Pierre Chaveau and Jean-Pierre Olivier Sardan (2000): ”Introduction.Les courtiers entre développement et l’Etat,” chapter 1 in Bierschenk, Thomas, Jean-Pierre Chaveau and Jean-Pierre Olivier Sardan (eds.):Courtiers en Développement. Paris:Karthala.

Blair, Harry (2000): “Participation and Accountability at the Periphery: Democratic Local Governance in Six Countries”. World Development 28 (1): 21-39. London: Elsevier Science Ltd.

Blundo, Giorgio (1995): ”Les courtiers du développement en milieu rural sénégalais”, Cahiers d’Études africaines 137 (1):73-99. Paris: EHESS.

Boone, Cathrine (1998): « State Building in the African Countryside :Structure and Politics at the Grassroots » The Journal of Development Studies 34 (4) :1-31. London :Franck Cass.

Bourdieu, Pierre (1977):Outline of a Theory of Practice. Cambridge :Cambridge University Press.

Bouju, Jacky(2000): «Clientelisme, gouvernance démocratique et corruption à Mopti ». Autrepart, 14:143-163.

Bratton, Michael and Nicholas van de Walle (1997): Democratic Experiments in Africa. Regime

Braathen, Einar, Morten Bøås and Gjermund Sæther (2000):”Ethnicity Kills? Social Struggles for Power, Resources and Identities in the Neo-Patrimonial State,”chapter 1 in Braathen, Einar, Morten Bøås,and Gjermund Sæther (eds.): Ethnicity Kills? The Politics of War, Peace and Ethnicity in Sub- Saharan Africa. London: Macmillan.

Cameron, Robert (2003):”Local Government in South Africa: Political Decentralisation, Party Centralisation, Corruption and Maladministration” in Hadenius (ed.): Decentralisation and Democratic Governance. Experiences from India, Bolivia and South Africa. Stockholm:Elanders Gotab.

CCC Bafoulabé (2001):Diagnostic de la commune rurale de Kontela. Centre Conseil Communal à Bafoulabé. Bafoulabé: CCC/République du Mali.

Chabal and Daloz (1999): Africa Works. Disorder as a Political Instrument. Oxford:James Currey.

Chavez, Daniel (2002):“Porto Allegre, Brazil: A new, sustainable and replicable model of participatory and democratic governance? Paper submitted to the conference on Local politics and Democratisation in developing countries, Oslo, October 21-22, 2002.(reading date:28.02.03)[online]URL:http://www.sum.uio.no/research/ democracy/network/Conference papers /Chaves1.pdf Conference paper.

Coll, Jerôme (1997): « Des dynamiques villageoises au service d’une « démocratie décentralisée. Le cas de Mali-Sud », APAD Bulletin 14:35-59. Lit Verlag, Hamburg.

Collier (1993): “The Comparative Method” in Finifter, A.W (ed.): Political science: The state of the Discipline II:105-119. American Political Science Association 129

Commune de Tambaga (2001): Plan Triennal de Developpement de la commune de Tambaga, 2001- 2003. Kita: Commune de Tambaga.

Councilor, Tambaga (2004.04.30) Questions [e-mail to Siri Bjerkreim Hellevik]- accessible from [email protected]

Cox, Kevin R. (ed.) (1997): Spaces of Globalization: Reasserting the Power of the Local. New York: Guilford.

Crook, Richard and Manor, James (1998): Decentralisation in South East Asia and West Africa. Participation, accountability and performance. Cambridge: Cambridge University Press.

Crook, Richard and Sverisson, Alan Sturla (2001):“Decentralisation and Poverty-Alleviation in Developing Countries: a comparative analysis, or is West-Bengal unique?” (reading date:10.03.03) [online]URL: http://www.ids.ac.uk/ids/bookshop/ wp/wp130.pdf IDS working paper:130

Davies, John Uniack (2000): “Classrooms of Democracy? The Educational Prospects of Malian Civil Society” chapter 17 in Bingen, R. James, David Robinson, and John M. Staatz (eds.) Democracy and Development in Mali. Michigan: Michigan State University Press.

De Langen, Maiken (2003):Réponses complémentaires à une même question: La décentralisation au Mali à travers l’experience du lotissement de Douentza. Unpublished paper. Not to be quoted without permission from the author.

Demos (2004): Executive Report from 1st Round Study of the Problems and Options of Indonesian Democratisation.(reading date:25.03.04) [online]URL:http://www.sum.uio.no/ research/democracy/ network/andrepapers/2Demos%201st%20Round%20Executive%20report.pdf Jakarta:Demos

Diallo, Toumani Djimé (1995): « La Décentralisation? Rendre son âme au Mali profound ». Le Democrate Malien, edition speciale July. Bamako: Imprim color.

Fay, Claude (2000): “La decentralisation dans un Cercle” in Autrepart 14:121-142. Bondy: IRD.

Foumonyoh, Christopher (2001):”Francophone Africa in Flux. Democratization in fits and starts” Journal of Democracy 12 (3):37-50. Baltimore:The John Hopkins University Press.

Gill, Graeme (2000):The dynamics of democratization. New York: St. Martin’s Press.

Guindo, Timothée (1999): Nature, vulgarisation et Organisations Professionnelle Agricoles. Montpellier:Centre National des Etudes agricoles des regions chaudes:MA thesis.

Hanke, Stephanie (2001):Systemwechsel in Mali. Bedingungen und Perspektiven der Demokratiserung eines neopatrimonialen Systems. Hamburg: Institut für Afrika-Kunde im verbund Deutsches Übersee-Institut. Hamburg Beiträge zur Afrika-kunde: 64.

Harris, John, Kristian Stokke and Olle Törnquist (2004):”Introduction. The New Local Politics of Democratisation” chapter 1 in Harris, John, Kristian Stokke and Olle Törnquist (eds.): Politicising Democracy. Local Politics and Democratisation in Developing Countries. Forthcoming, London:Palgrave Macmillian

Haugsjå, Anita og Hølje (2002): Helse og Alfabetisering I Tambaga. Pilotprosjekt i Tambaga og Kokofata kommuner, Kita fylke, Mali. Project document to NORAD, June 2002. Not published. 130

Heller, Patrick (2001): “Moving the State: The Politics of Democratic Decentralization in Kerala, South Africa, and Porto Allegre” Politics and Society 29 (1):131-163. Thousand Oaks: Sage Publications.

Hellevik, Ottar (1991):Forskningsmetode i Sosiologi og Statsvitenskap. Oslo: Universitetsforlaget.

Hetland, Øivind (2000): Stat, samfunn og politikk: en studie av desentraliseringsformen i Mali. Oslo: Universitetet i Oslo, Institutt for samfunnsgeografi: Hovedoppgave (MA thesis).

Hopkins, Nicholas S. (1968): Popular Government in an African Town. Chicago: The University of Chicago Press.

L’Intelligent Jeune Afrique (June 6, 2004):”Les Municipales placent L’Adema et L’URD en tête (résulats complets)” (reading date 06.06.04) [online] URL:http://www.lintelligent.com/gabarits/ articleAFP_online.asp?art_cle=AFP20624lesmustelpm0

Johnson (2001): Local Democracy, Democratic Decentralisation and Rural Development: Theories, Challenges and Options for Policy. Development Policy Review 19(4): 521-533.

Kassibo (1997): “La Décentralisation au Mali: État des Lieux”. APAD Bulletin 14: 1-21. Lit Verlag, Hamburg.

(2002):”Participatory Management and Democratic Decentralization Management of the Samori Forest in Babye Commune, Mopti Region, Mali”. Conference paper to the Annual Conference of the International Association for the Study of Common Property, Victoria Falls, Zimbabwe, June. (reading date:26.03.03) [online]URL:dlc.dlib.indiana.edu/documents/dir0/00/ 00/ 08/43/dlc-00000843-00/kassibob210502.pdf Conference paper.

Koné, Soli (1997): “La décentralisation face à l’ordre ancien” in APAD Bulletin, no14:67-79. Hamburg. Lit Verlag.

Lange, Marie-France (1999):”Insoumission civile et défaillance étatique: les contradictions du processus démocratique malien” in Autrepart, 10:117-134. Bondy: IRD.

Le Marcis (1999):”Histoire, pouvoirs, pâture forestière et droits d’un groupe peul Baalinkoobe (Mali),” chapter 6 in Bourgeot, André (ed.) : Horizons Nomades en Afrique Sahélienne. Paris: Karthala.

Lemarchand, Réne (1998):” La Face Cachée de la Décentralisation: Réseaux, Clientèles et Capitale Social” in APAD Bulletin 16:9-17. Hamburg: Lit Verlag.

Lintz, Juan and Alfred Stepan (1996): Problems of Democratic Transition and Consolidation. Southern Europe, South America and Post-Communist Europe. Baltimore: John Hopkins Press.

Lippman, Hal and Barbara Lewis (1998): Democratic Decentralization in Mali: A work in progress. CDIE Impact Evaluation, no2, USAID. (reading date: 21.07.04)[online] Washington:USAID. URL:http://www.dec.org/pdf_docs/phaca905.pdf

Mamdani, Mahmood (1996): Citizens and Subjects. Contemporary Africa and the legacy of late colonialism. Princeton:Princeton University Press.

Massing, Andreas (1996a): “La commune de Mopti. Structures traditionelles, vie politique et decentralisation”(reading date:26.01.04)[online]URL:http://home.t-online.de/home/ 320033440512- 0001/downloads2/histmopt.pdf and from (2004, January 28): Casestudy of Mopti [e-mail to Siri B. Hellevik] [online]- accessible via e-mail:[email protected] 131

(1996b):“Relations entre structures socio-politiques traditionelles et modernes dans la commune de San. Démocratisation et décentralisation”. (reading date: 30.01.03)[online]URL:http://home.t-online.de/home/320033440512-0001/downloads2/histsan.pdf

(2000):”Decentralization Policy and Traditional Authority in Mali: San, Mopti and Djenné”.Paper abstract to the 43rd Annual Meeting of the African Studies Association, Nashville, USA, Nov. 16-19. (reading date:12.02.04) [online] URL:http:// home.t- online.de/home/320033440512-0001/downloads2/ malidece.pdf

Mathew, George with Anand Mathew (2003) : « India :Decentralisation and Local Governance. How Clientelism and Accountability Work » in Hadenius (ed.) Decentralisation and Democratic Governance. Experiences from India, Bolivia and South Africa. Stockholm:Elanders Gotab.

Mbembe, Achille (2001): On the Postcolony. Los Angeles: California University Press.

Mc Vey (2000): Money and Power in Provincial Thailand. Copenhagen:NIAS.

Médard, Jean-Francois (1976):”Le rapport de clientele” in Revue Francaise de science politique, 26 (1):103-131.

(1991) : « Etatisation et desétatisation en Afrique Noire » in Médard, Jean Francois(ed.) :Etats d’Afrique Noire :Formations, mécanismes et crises. Paris :Karthala.

(1996): “Patrimonialism, Neo-patrimonialism and the study of the Post- colonial State in Sub-Saharan Africa” in Secher Marcussen, Henrik (ed.): Improved Natural Resource Managent - the role of formal organisations and informal networks and institutions. Roskilde: IDS.

Mission de Décentralisation(MDD) (1997): Aperçu de la Politique de décentralisation administrative. Bamako: République du Mali.

(document without date) :Cartographie des Communes Rurales en République du Mali. Bamako :République du Mali

Mission de Décentralisation et Réformes Institutionelles (MDRI) (1998):La Commune en questions.... Bamako: République du Mali.

Mkwandawire, Thandika (1994):”Adjustment, Political Conditionality and democratisation in Africa” in Cornia and Helleiners (eds.):From Adjustment to Development in Africa:conflict, controversy, convergence, consensus? London: Macmillian.

Mohan and Stokke (2000): Participatory Development and Empowerment: The dangers of Localism. Third World Quarterly, 21, 2: 247-268. London: Carfax Publishing Company.

Ndegwa, Stephen N. and Helene Grandvoinnet (2002): Decentralization in Africa: A Stocktaking Survey. Africa Region Working Paper Series, no 40. (reading date:17.02.03) [online]URL: http://www.ndegwa.com/ images/Africa_Region _WP_40.pdf

N’Diaye, Bokar (1995): Les castes au Mali. Paris: Présence Africaine

Nijenhuis, Karen (2003): “Does decentralisation serve everyone? Struggle for Power in a Malian village”. European Journal of Development Research 15 (2):67-92.

Olokushi, Adelabayo (1998):The elusive Prince of Denmark. Uppsala:Nordiska Afrikainstitutet. Research Report: 104. 132

Olowu, Dele (2001):Decentralization Policies and Practices under Structural Adjustment and Democratization in Africa. New York:UNRISD. Democracy, Governance and Human Rights Programme Paper: 4 (July).

Olowu, Dele and James S. Wunsch (2004): Local governance in Africa. The challenges of Democratic Decentralization. Boulder: Lynne Rienner Publishers.

Owusu, Maxwell (1992): “Democracy and Africa. A View from the Village” The Journal of Modern African Studies. (reading date:15.01.01)[online] URL:http://kinks.jstor.org/sici=0022- 278X%28199209%2930%3A3%3C369%3 ADAA-AV%3E2.0.CO%3B2-J

Ragin, Charles C. (1994): “Introduction to Qualitative Comparative Analysis” in Janoski, Thomas, and Alexander M Hicks (eds.): The Comparative Political Economy of the Welfare State: New Methodologies and Approaches. Cambridge: Cambridge University Press

Rawson, David (2000): “Dimensions of Decentralization in Mali” chapter 16 in Bingen, R. James David Robinson, and John M.Staatz (eds.) Democracy and Development in Mali. Michigan: Michigan State University Press.

Reigstad, Ingeborg Brenne (2002): Fri til å velge? En studie av hvordan afrikanske samfunnsforskere utfordrer vestlig liberaldemokratidiskurs. Oslo:Universitetet i Oslo, Institutt for Statsvitenskap. Hovedoppgave (MA thesis).

Reno, William (1998): Warlord Politics and African States. Boulder: Lynne Rienner Publishers.

République du Mali (1997): Lois et Décrets de la Décentralisation. Bamako:République du Mali.

(2001): Decision no 0040, 2001 MATCL-S:G ANICT. Fixant la liste des Investissements Eligibles aux fonds de l’agence nationale des investissements des collectivités territoriales. Bamako: Le République du Mali.

(2002):Plan de Développement Socio-Sanitaire du Cercle de Bafoulabé 2003- 2007. Bafoulabe:Service Socio-Sanitaire, Direction Régionale de la Santé, Région de Kayes.

(2003) :Fiche Ville Mopti-Sévare. Projet de developpement urbain et decentralisation, vol 2. Bamako:Ministère des domains de l’Etat et des affaires foncières et de l’Habitat/Banque Mondiale/Agetipe.

(document not dated) :Mouvement Synergique de Developpement du Cercle du Bafoulabé (M.S.D.B). Projet Cadre de Reference. Bafoulabé : Comité Local de Développement, Cercle de Bafoulabé République du Mali.

Ribot, Jesse (2002): “African Decentralization. Local Actors, Powers and Accountability”. New York:UNRISD. Democracy, Governance and Human Rights Paper: 8.

Rubin, Herbert J. and Rubin, Irene S. (1995): Qualitative Interviewing. The Art of Hearing Data. Thousand Oaks: Sage Publications.

Rudebeck, Lars (2001): On Democracy’s Sustainability. Transition in Guinea Bissau. Göteborg: SIDA. SIDA Studies: 4.

Sandbrook, Richard (2000): Closing the Circle. Democratization and Development in Africa. London: Zed Books Ltd. 133

Sawadogo, Raogo Antoine: L’État africain face à la decentralisation. La chaussure sur la tête. Paris:Karthala.

Schumpeter, Joseph A. (1987): Capitalism, Socialism and Democracy. London: Unwin Paperbacks.

Seely, (2001): “A political analysis of decentralization: coopting the Tuareg threat in Mali”. Journal of Modern African Studies 39 (3): 499-524. Cambridge:Cambridge University Press.

Sidel, John T. (1999):Capital, Coercion and Crime. Bossism in the Phillippines. Stanford:Stanford University Press.

Sundstøl-Eriksen, Stein, Jon Naustdalslid and Arild Schou (1999):Decentralisation from Above. A study of local government in Botswana, Ghana, Tanzania and Zimbabwe. Oslo: NIBR. NIBR’s Pluss Series:4.

Sundstøl-Eriksen, Stein (2000): Close Links and Blurred Boundaries: Council and Commmunity in a Tanzanian and a Zimbabwean District. Oslo:Universitetet i Oslo, Institutt for Statsvitenskap. Doktorgadsavhandling (Doctoral Dissertation).

State employee, Mopti (2004.03.21) Reponses aux questions sur le transfer des competences [e-mail til Siri Bjerkreim Hellevik]- accessible via e-mail:[email protected]

Stokke, Kristian and Olle Törnquist (2001):”Local Politics and Democratisation in Developing Countries”. Project Proposal to Norges Forskningsråd, June. Unpublished.

Stokke, Kristian (2002): Habitus, capital and fields. Conceptualizing the capacity of local actors in local politics. Paper submitted to the conference on Local politics and Democratisation in developing countries, Oslo, October 21-22, 2002. (reading date:18.02.03) [omline)URL: http://www.sum.uio.no/research/ democracy/network/Conferencepapers/ Stokke.pdf Conference Paper.

Sy, Ousmane (ed.) (1998): Décentralisation et reformes institutionnels au Mali. Cadre d’une nouvelle dynamique de democratization et de développement. Bamako: MDRI.

Tag, Sylvia (1994):Paysans, Etat et Démocratisation au Mali. Enquête en milieu rural. Hamburg: Institut für Afrikakunde. Hamburg African Studies/Etudes Africaines Hamburgoises : 4.

Tandon, Yash (1995): ”Hvorfor den vestlige demokratidiskursen har spilt fallitt in Afrika?” in Patrick Eckløf (ed.): Demokrati I Afrika. Fellesrådets Årbok 95/96. Oslo: Gazette bok.

Tavares, Pierre Franklin (2004): “Désintegration des souverainetés nationals. Pourquoi tous ces coups d’Etat en Afrique?” Le Monde Diplomatique, January:16-17. (reading date: 06.07.04) [online]URL:http://www.monde-diplomatique.fr/2004/01/TAVARES/10927?var_ recherche= Coups + Afrique

Tendler, Judith (1997): Good government in the Tropics. Baltimore/London: John Hopkins University Press.

Tharakan, Michael (2002):”Forces of Change in Local Politics: lessons from the Kerala Experience” Paper submitted to the conference on Local politics and Democratisation in developing countries, Oslo, October 21-22, 2002. (reading date: 03.02.03) [online)URL: http://www.sum.uio.no/research/democracy/network/ Conferencepapers/ Tharakan.pdf Conference Paper. 134

Therkildsen, Ole (2000):”Demokratisk? Desentralisering?” Den Ny Verden 4:5-15. København: Center for Udviklingsforskning.

Tostensen, Arne, Inge Tvedten and Mariken Vaa (2001): “The Urban Crisis, Governance and Associational Life,” chapter 1 in Tostensen, Arne, Inge Tvedten and Mariken Vaa (eds.): Associational life in African cities. Popular responses to the Urban Crisis. Uppsala: Nordiska Afrikainstitutet.

(1999):Politics and Development. A critical introduction. London:Sage.

Törnquist, Olle (2002): Popular Development and Democracy. Case Studies with rural dimensions in the Philippines, Indonesia and Kerala. Oslo: Senter for Utvikling og Miljø.

(2003):Assessing Democratisation: Background, Design and Questionnaire, Collection of Project Papers. Oslo: paper presented at conference, Uppsala March 12-13, 2003. Conference paper.

(2004):”The Political Deficit of Substantial Democratisation” in Harriss, John, Kristian Stokke and Olle Törnquist (eds.):Politicising Democracy. Local Politics and Democratisation in Developing Countries. Forthcoming, London:Palgrave Macmillian

UNDP (2002): “Deepening Democracy by Tackling Democratic Deficits” in Human development Report 2002: Deepening Democracy in a Fragmented World. Oxford:Oxford University Press.

(2003): « Evaluation d’effet appliqué au theme de la décentralisation ». Rapport final, Mission d’évalution de l’Effet 3 du Cadre de Résultats Stratégiques du PNUD. Bamako:UNDP.

(2004):Human Development Report 2004: Cultural Liberty in Today’s Diverse World. (reading date:15.07.04) [online] URL:http://hdr.undp.org/reports/ global/2004/pdf/hdr04_HDI.pdf

Vedeld (2002):”State Law versus Village Law: law as an exclusion principle under customary tenure regimes”. Paper presented at Presented at "Crossing Boundaries," the seventh annual conference of the International Association for the Study of Common Property, Vancouver, British Columbia, Canada, June 10-14, 1998. [online] (reading date:12.05.04) URL:dlc. dlib. indiana.edu/ document/s/ dir0/00/00/01/79/dlc-00000179-00/vedeld.pdf Conference paper.

(2003):”Democratic Decentralisation and Poverty Reduction:Exploring the Linkages” in Forum for Development Studies, no.2, Vol.30:159-204. Oslo:NUPI.

Vengroff, Richard (1993): “Governance and the transition to Democracy: Political Parties and the Party system in Mali”. The Journal of Modern African Studies, vol. 31, 4:531-562. Cambridge: Cambridge University Press.

Villadsen, Søren Taastrup (1999): “Decentralisation and Good Governance Reforms in Africa” in Villadsen, Søren (ed.) Good Governance and Decentralization. Public Sector Reforms in Developing Countries. Taastrup:Nordic Consulting Group.

Walle, Nicolas van de (2001):”Rejoinder to Inge Amundsen and Stein Sundstøl Eriksen” in Forum for Development Studies 28(1):69-75. Oslo:NUPI.

(2002):”Elections without Democracy. Africa’ s Range of Regimes” in Journal of Democracy 13 (2):66-80. (reading date 20.01.02) [online]URL: http://swproxy.swetswise.com/proxy?url=http%3A%2F%2Fmuse.jhu.edu%2Fjournals%2Fjournal_o f_democracy%2Fv013%2F13.2van-de-walle.html&ts=1092313963307&cs=429483 2934 135

Weber, Max (1997):Makt og byråkrati. Sentrale verker av ”byråkratiteoriens far” og den betydeligste samfunnsforsker i nyere tid. Oslo:Gyldendal.

World Bank (1992): Governance and Development. Washington D. C: World Bank

(1997): The state in a Changing World. World Development Report. Oxford: Oxford University Press.

Yin (1984): Case Studies Research: Design and Methods. Beverly Hills: Sage.

Websites:

Decentralization: Greater involvement in decision-making processes leads to greater potential for effective, sustainable and equitable development (reading date 19.03.03) [online] URL: http://www. undp.org/ governance/decentralization.htm

Decentralization: What, Why, Where (reading date:07.04.04)[online](source a): URL:http://www1. worldbank.org/ publicsector/ decentralization/what.htm

Etat de Decentralisation au Mali (reading date:10.09.03) [online]: URL: http://www.pdm- net.org/Newsite/french/decentralisation/mali/etat_decent_mali.pdf

Fiscal Decentralization (reading date 12.06.04) [online] URL:http://www.fao.org/DOCREP/ 005/Y4256E/y4256e05.htm.

Governance for Sustainable Development. A UNDP Policy Document. Good Governance and Sustainable Development (reading date:23.04.04) [online]:URL:http://magnet. undp.org/ policy/chapter1.htm#b

Intergovernmental Fiscal Relations (reading date:07.04.04) [online] Source B : URL:http://www1. worldbank.org/publicsector/ Decentralization Sub NationalEconomics/fiscal.htm

Mali Data Profile (reading date 12.03.04)[online](source b) URL:http://devdata.worldbank.org/ external/CPProfile.asp?CCODE=MLI&PTYPE=CP

Mali 2004 - Annual Report of Press Freedom. Reporters sans frontières. [online] (reading date: 17.08.04) URL:http://www.rsf.fr/article.php3?id_article=10182

Table of Independent Countries 2003 [online] (reading date 12.08.04) URL:http://www. freedomhouse.org/research/freeworld/2003/table.pdf

Interviews:

In Kontela:

Municipal employee(2003) interview. October 6

Municipal employee (2003) interview. October 6

Councilor (2003) interview. October 9

Peasant (2003)interview. October 10

Councilor (2003) interview. October 13 136

Councilor (2003) interview. October 12

Councilor (2003) interview. October 8

Councilor (2003) interview. October 13

Councilor (2003) interview. October 15

Woman (2003) interview. October 14

Anonymous (2003) interview. October 17

Councilor (2003) interview. October 13

State Employee (2003) interview. October 14

Councilor (2003) interview. October 14

Woman (2003) interview. October 15

Woman (2003) interview. October 17

Woman (2003) interview. October 15

Woman (2003) interview. October 15

Woman (2003) interview. October 15

Peasant (2003) interview. October 15

State employee(2003) interview October 14

State Employee (2003) interview. October 11

Councilor (2003) interview October 9

Councilor (2003)interview. October 14

Municipal employee (2003) interview. October 16

Peasant (2003) interview. October 14

Municipal employee (2003) interview. October 23

10 out of 13 councilors in Kontela’s municipal council (2003). Group interview. October 9.

In Tambaga:

Teacher (2003) interview. October 29

Municipal employee (2003) interview October 28

Councilor(2003) interview. October 28

Councilor(2003) interview. October 28 137

Councilor (2003) interview. October 27 and 28

Village chief (2003) interview. October 29

Woman (2003) interview. October 29

Woman (2003) interview. October 29

Village chief (2003) interview. October 30

Municipal employee (2003) interview. October 28

5 teachers (2003). Group interview. November 1.

State employee (2003) interview. October 31

Tomora:

Mamadou Sissoko (2003), interview. October 18.

Mopti:

NGO worker interview. (Date left out, because informant wanted to)

NGO worker (2003) interview. November 10

Councilor (2003) interview. November 11

State employee (2003) interview. November 11

NGO worker (2003) interview. November 11

State employee (2003) interview. November 11(together with the one listed above)

State employee (2003) interview. November 12

NGO worker (2003) interview. November 12

Anonymous (2003) interview. November 12

NGO worker (2003) interview. November 13 (together with the one listed above)

NGO worker (2003) interview. November 13 (interview with the two listed below)

NGO worker (2003) interview. November 13

NGO worker (2003) interview. November 13

Bamako:

Kassibo, Bréhima (2003), interview. September 23.

Danté, Gaoussou (2003) interview. September 24

Beridogo, Bréhima (2003) interview. November 18 138

Sénou Touré (2003) interview. November 18

Konaté, Yacouba (2003) interview. November 19

Ouane, Mamadou (2003) interview. November 21

Diakité, Youssouf (2003) interview. November 24

Soumounou, Salif (2003) interview. November 24

Cissé, Fatomata (2003) interview. November 24

Norway:

Amundsen, Petter (2004) phone interview. February 2.

Appendix 1: Professional titles of informants who have not been anonymized Brehima Kassibo-Research director, Institut des Sciences Humaines, Bamako Brehima Beridogo-anthropologist, Faculté des lettres, langues, arts, et sciences humaines, Université du Mali, Bamako Yacouba Konaté- USAID Democratic Governance program Mali Senou Touré- Councilor, Direction Nationale de Collectivités Territoriales under Le Ministère de l’Administration des Collectivités Territoriales. Youssouf Diakité- Permanent secretary, Association des Municipalités du Mali Salif Soumonou-Agence Nationale des Collectivités Territoriales Mamadou Ouane- Coordinator of“Projet de developpement urbain et decentralisation” under Ministère des domains de l’Etat et des affaires foncières et de l’Habitat Fatoumata Cissé- Advisor, Norwegian Church Aid Gaoussou Danté- responsible de la cellule Structuration, Appui aux Collectivités Décentralisées pour un Développement Participatif (ACODEP), UNDP Mali. Petter Amundsen- former missionary in Kontela for Normisjon

Appendix 2: Framework of Substantial Democratization (Törnquist, 2002, 2003)

Figure 1:Substantial democracy/democratization (Törnquist 2003)

bypass Actor's Instrumental Aims bypass

I. R&I re. Law, Judiciary, Citizenship II. R&I re. Government, Public Administration, and Human Rights Representation and Accountability

III. R&I re. Civil Society

incl. instances of direct dem o cracy & self -m anagem ent .

IV. Actors' Capacity to Use and Promote Rights & Institutions The Actors 139

Key Factors Conditioning Actors' Capacity

opportunity structure; sources of power; transformation of power into authority, legitimacy, honour; guiding values/perspectives)

Figure 2: The political terrain for action (Törnquist 2002:41)

Mobilisation

Incorporation Alternative Patronage Integration Ideas/interests

Populism<—–>Clientelism Incorporation<––>Integration Networks<–>Organised integration

Single issues/

Specific interests Populism<—–>Clientelism Incorporation<––>Integration Networks<–>Organised integration

Ideology/

Collective interests Populism<—–>Clientelism Incorporation<––>Integration Networks<–>Organised integration

Values/Communal

Loyalties Populism<—–>Clientelism Incorporation<––>Integration Networks<–>Organised integration

140

Figure 3: Politicisation of interests and ideas – and people (Törnquist 2002:36) NB! Civil-political societies, which are included in this model from 2002 have later been left out by the author, and are thus not discussed in this thesis.

Appendix 3: A summary of the assessment of political actors’ capacity in section 7.2:

MOPTI TAMBAGA KONTELA STATE Link activities? No, only local state No, only local state No, only local state Politicization Single and collective No politicization No politicization interests Mobilization Incorporation Integration Integration Joint strategy Towards clientelism Towards integration Towards integration

Relation regarding Abusers (promoters in User and abuser Abusers (promoters in Rights/Institutions some instances) (promoters in some some instances) instances) VILLAGE CHIEFS Link activities? Yes Yes Yes Politicization Single and coll. Single and coll, interests Single and coll. interests and communal and communal loyalties interests and communal 141

loyalties loyalties Mobilization Incorporation Incorporation Incorporation

Joint strategy Clientelism Integration Integration Relation R/I Mostly abusers, Mostly abusers, promoters Mostly abusers, promoters when when representing the promoters when representing the city village representing the village PARTIES

Link activities? Yes Yes Yes

Politicization Single issues/interests Single issues and some Single issues and some communal loyalties communal loyalties Mobilization Incorporation Incorporation Incorporation Joint strategy Towards clientelism Towards clientelism Towards clientelism Relation R/I Users and abusers Users and abusers Users and abusers COUNCILORS Link activities? Yes Yes Yes Politicization Single issues, a few Single issues, communal Single issues, a few have collective interests loyalties have collective interests & communal loyalties Mobilization Incorporation Incorporation Incorporation

Joint strategy Closest to clientelism Closest to clientelism Closest to clientelism

Relation R/I Users and abusers Users and abusers Users, some promoters

CIVIL SOCIETY

Link activities? Yes No No

Politicization Collective interests Collective interests Collective interests

Mobilization Integration Integration Integration

Joint strategy Between networks and Between networks and Between networks and integration integration integration

Relation R/I Pro-democrats Bypassing No relation